CN118609368A - A method, device and medium for intelligent control of tunnel traffic based on holographic perception - Google Patents

Abstract

The present application discloses a method, device and medium for intelligent control of tunnel traffic based on holographic perception, which belongs to the technical field of traffic control systems. The method includes: obtaining multi-source vehicle information in the tunnel to be controlled through multiple data sources, and fusing the multi-source vehicle information corresponding to each vehicle based on the Kalman filter algorithm; predicting the vehicle behavior in the tunnel to be controlled based on the historical traffic information in the tunnel to be controlled and the fused multi-dimensional information of the vehicle; determining the vehicle motion trajectory according to the multi-dimensional information of the vehicle, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle driving data and vehicle motion trajectory in the multi-dimensional information of the vehicle, and predicting the traffic safety risk points in the tunnel to be controlled based on the vehicle behavior and traffic situation; adjusting the tunnel control strategy based on the traffic safety risk points, so as to realize intelligent control of tunnel traffic according to the optimized tunnel control strategy.

Description

A method, device and medium for intelligent control of tunnel traffic based on holographic perception

Technical Field

The present application relates to the technical field of traffic control systems, and in particular to a method, device and medium for intelligent control of tunnel traffic based on holographic perception.

Background Art

At present, with the continuous development of the transportation industry, tunnel traffic, as an important part of the modern transportation system, has received increasing attention for its safety and efficiency. However, due to the closedness and particularity of tunnels, the identification and control of traffic safety risks are particularly important. At present, the identification of traffic safety risks in tunnels mainly relies on video monitoring and intelligent recognition technology, which captures vehicle images through cameras installed in the tunnel and uses image processing and machine learning algorithms to identify vehicles.

The existing video monitoring and intelligent recognition systems mainly focus on the recognition of each vehicle, such as vehicle model recognition, license plate recognition, etc., and their recognition dimensions are relatively single. The advantage of this method is that it can capture the basic information of vehicles in the tunnel in real time and accurately, providing certain data support for traffic management and safety monitoring.

However, this single-dimensional identification method also has obvious limitations. It lacks in-depth analysis of the temporal operation continuity of vehicles in the tunnel and comprehensive consideration of the spatial operation continuity between vehicles. Specifically, existing systems can often only provide a static "snapshot" of the vehicle at a certain point in time, but cannot reflect the dynamic behavior characteristics of the vehicle in the tunnel, such as acceleration, deceleration, lane changing behavior, etc. At the same time, these systems also fail to fully consider the relative position and speed relationship between vehicles, making it difficult to fully evaluate the traffic flow characteristics and potential safety risks in the tunnel. The existing tunnel traffic safety risk identification system has certain limitations in preventing traffic accidents, optimizing traffic flow and improving tunnel traffic efficiency.

Summary of the invention

The embodiments of the present application provide a method, device and medium for intelligent control of tunnel traffic based on holographic perception, which is used to solve the technical problems that the existing traffic safety risk identification mainly relies on video monitoring and intelligent identification, the identification dimension is relatively single, and there is a lack of analysis of the temporal operation continuity of vehicles in the tunnel and the spatial operation continuity between vehicles.

On the one hand, the embodiment of the present application provides a method for intelligent management and control of tunnel traffic based on holographic perception, including:

Obtain multi-source vehicle information in the tunnel to be controlled through multiple data sources, and fuse the multi-source vehicle information corresponding to each vehicle based on the Kalman filter algorithm;

Predicting vehicle behavior in the tunnel to be controlled based on historical traffic information in the tunnel to be controlled and the integrated multi-dimensional vehicle information;

Determine the vehicle motion trajectory according to the vehicle multi-dimensional information, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle driving data and the vehicle motion trajectory in the vehicle multi-dimensional information, and predict the traffic safety risk points in the tunnel to be controlled according to the vehicle behavior and the traffic situation;

Based on the traffic safety risk points, the tunnel control strategy is adjusted to achieve intelligent control of tunnel traffic according to the optimized tunnel control strategy.

In one implementation of the present application, the multi-source vehicle information corresponding to each vehicle is fused based on the Kalman filter algorithm, specifically including:

Cleaning the multi-source vehicle information to remove abnormal data and duplicate data in the multi-source vehicle information, and calibrating the cleaned multi-source vehicle information;

Performing time synchronization processing on the cleaned and calibrated multi-source vehicle information to obtain standard multi-source vehicle information, and extracting vehicle features associated with abnormal traffic events from the standard multi-source vehicle information;

The vehicle features are fused by a Kalman filter algorithm to obtain fused multi-dimensional vehicle information; wherein the multi-dimensional vehicle information includes at least: vehicle position, vehicle speed, acceleration and vehicle type;

Determine the measurement noise of the perception device corresponding to the vehicle multi-dimensional information, and based on the measurement noise, determine the compensation coefficient corresponding to the vehicle multi-dimensional information, so as to compensate the vehicle multi-dimensional information according to the compensation coefficient to obtain compensated vehicle multi-dimensional information.

In one implementation of the present application, predicting the vehicle behavior in the tunnel to be controlled based on the historical traffic information in the tunnel to be controlled and the fused multi-dimensional vehicle information specifically includes:

Determine the tunnel traffic characteristics and vehicle movement rules corresponding to the tunnel to be controlled based on the historical traffic information of the tunnel to be controlled, and extract the tunnel impact factor corresponding to the tunnel to be controlled from the historical traffic information;

According to the tunnel traffic characteristics and vehicle operation rules, a vehicle state model corresponding to the tunnel to be controlled is established, and the fused vehicle multi-dimensional information is associated with the vehicle state model to form an observation model;

Based on the vehicle state model and the observation model, and according to the tunnel influencing factor, the vehicle multi-dimensional information is analyzed to obtain the vehicle state in the tunnel to be controlled, so as to predict the vehicle behavior in the tunnel to be controlled based on the vehicle state.

In one implementation of the present application, determining the vehicle motion trajectory according to the vehicle multi-dimensional information, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle driving data and the vehicle motion trajectory in the vehicle multi-dimensional information, specifically includes:

Determine a sampling position and a sampling time corresponding to the multi-dimensional information of the vehicle to generate a data sampling trajectory corresponding to the multi-dimensional information of the vehicle;

Based on the data sampling trajectory, fitting the vehicle position and speed in the vehicle multi-dimensional information to determine the vehicle motion trajectory of each vehicle in the tunnel to be controlled;

The traffic flow parameters in the tunnel to be controlled are calculated according to the multi-dimensional information of the vehicle and the movement trajectory of the vehicle, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the traffic flow parameters and the vehicle behavior; wherein the traffic flow parameters include vehicle volume, average vehicle speed, vehicle density and road occupancy rate.

In one implementation of the present application, predicting the traffic safety risk points in the tunnel to be controlled based on the vehicle behavior and the traffic situation specifically includes:

Inputting the vehicle behavior into the observation model to determine the vehicle behavior and reward value corresponding to each vehicle in the tunnel to be controlled at the next moment;

Determine the descent gradient corresponding to each vehicle in the tunnel to be controlled at the next moment according to the traffic situation;

The vehicle behavior, the reward value and the descent gradient corresponding to each vehicle at the next moment are input into a pre-trained safety identification model to determine the risky vehicles with potential risks in the tunnel to be controlled and the vehicle positions corresponding to the risky vehicles, and based on the vehicle positions, determine the traffic safety risk points in the tunnel to be controlled.

In one implementation of the present application, the tunnel control strategy is adjusted based on the traffic safety risk point, specifically including:

According to the traffic safety risk point, the corresponding area to be controlled in the tunnel to be controlled is determined, and multi-dimensional information of vehicles and the actual situation of the road section in the area to be controlled are obtained; wherein the actual situation of the road section includes the existence of obstacles, traffic accidents, and the intrusion of pedestrians and non-motor vehicles;

Determine the influence coefficient of the road section actual condition on each vehicle in the area to be controlled, and determine the corresponding vehicle behavior in combination with the influence coefficient and according to the multi-dimensional information of the vehicles in the area to be controlled;

According to the vehicle behavior in the area to be controlled, the risk event type is determined to generate a tunnel control instruction corresponding to the risk event type, and according to the tunnel control instruction, the traffic flow in the area to be controlled is adjusted to realize intelligent control of tunnel traffic.

In one implementation of the present application, after the tunnel control strategy is adjusted based on the traffic safety risk point to realize intelligent control of tunnel traffic according to the optimized tunnel control strategy, the method further includes:

Obtaining traffic feedback data in the area to be controlled in real time through the sensing device of the area to be controlled, and determining the execution effect corresponding to the tunnel control instruction according to the traffic feedback data;

According to the deviation between the execution effect and the expected tunnel control effect, the loss function corresponding to the tunnel control instruction is determined, so as to optimize the tunnel control instruction according to the loss function.

In one implementation of the present application, the acquiring of multi-source vehicle information in the tunnel to be controlled through multiple data sources specifically includes:

Determine a tunnel to be controlled, obtain historical traffic information corresponding to the tunnel to be controlled, and determine, based on the historical traffic information, an abnormal traffic event in the tunnel to be controlled and an abnormal event type corresponding to the abnormal traffic event;

Based on the abnormal event type, the data type corresponding to the potential abnormal data is determined, and the sensing device corresponding to each potential abnormal data is determined, so as to obtain multi-source vehicle information through the multiple sensing devices set in the tunnel to be controlled.

On the other hand, the embodiment of the present application further provides a tunnel traffic intelligent management and control device based on holographic perception, the device comprising:

at least one processor;

and, a memory communicatively coupled to the at least one processor;

Among them, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned intelligent tunnel traffic control method based on holographic perception.

On the other hand, an embodiment of the present application further provides a non-volatile computer storage medium storing computer executable instructions, which, when executed, implements a method for intelligent tunnel traffic control based on holographic perception as described above.

The embodiments of the present application provide a method, device and medium for intelligent control of tunnel traffic based on holographic perception, which at least have the following beneficial effects:

By acquiring multi-source vehicle information in the tunnel through multiple data sources and fusing this information using the Kalman filter algorithm, the accuracy and completeness of vehicle information can be significantly improved; based on the historical traffic information in the tunnel and the fused multi-dimensional vehicle information, the vehicle behavior in the tunnel can be predicted, which helps to identify potential traffic problems in advance; by determining the vehicle movement trajectory and combining the driving data in the vehicle multi-dimensional information, the traffic situation in the tunnel can be perceived in real time, and abnormal situations such as traffic congestion and accidents can be discovered in time; based on vehicle behavior and traffic situation, the traffic safety risk points in the tunnel can be predicted, which helps to prevent traffic accidents and improve the safety of tunnel driving; based on the predicted traffic safety risk points, the tunnel control strategy is adjusted to achieve intelligent control of tunnel traffic, which can not only improve the tunnel's traffic efficiency, but also respond quickly in emergency situations and reduce accident losses.

BRIEF DESCRIPTION OF THE DRAWINGS

The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

FIG1 is a schematic diagram of a flow chart of a method for intelligent tunnel traffic control based on holographic perception provided in an embodiment of the present application;

FIG2 is a schematic diagram of the internal structure of an intelligent tunnel traffic control device based on holographic perception provided in an embodiment of the present application.

DETAILED DESCRIPTION

In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

The embodiments of the present application provide a method, device and medium for intelligent control of tunnel traffic based on holographic perception, which solves the technical problems that the existing traffic safety risk identification mainly relies on video monitoring and intelligent identification, the identification dimension is relatively single, and there is a lack of analysis of the temporal operation continuity of vehicles in the tunnel and the spatial operation continuity between vehicles.

The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

FIG1 is a flow chart of an intelligent tunnel traffic control method based on holographic perception provided in an embodiment of the present application.

The analysis method involved in the embodiments of the present application can be implemented by a terminal device or a server, and the present application does not impose any special restrictions on this. For the convenience of understanding and description, the following embodiments are described in detail by taking a server as an example.

It should be noted that the server may be a single device or a system consisting of multiple devices, that is, a distributed server, and this application does not make any specific limitation on this.

As shown in FIG1 , an embodiment of the present application provides a method for intelligent tunnel traffic control based on holographic perception, including:

101. Obtain multi-source vehicle information in the tunnel to be controlled through multiple data sources, and fuse the multi-source vehicle information corresponding to each vehicle based on the Kalman filter algorithm.

In order to realize intelligent control of tunnel traffic, the server first determines the tunnels to be controlled that have control needs, and obtains various environmental factors such as sound, light, temperature, humidity, smoke, etc. in the tunnel through various sensors installed in the tunnel to be controlled, and obtains the speed of vehicles in the tunnel to be controlled through radar speed meters. Communication is established with vehicles entering the tunnel to be controlled through roadside equipment to obtain the vehicle's location information through the vehicle's own GPS positioning system, and video frame images of vehicles entering the tunnel to be controlled are obtained through cameras installed in the tunnel to be controlled.

Specifically, the server determines the tunnel to be controlled, obtains the historical traffic information corresponding to the tunnel to be controlled, and determines the abnormal traffic events in the tunnel to be controlled and the abnormal event types corresponding to the abnormal traffic events based on the historical traffic information; based on the abnormal event type, determines the data type corresponding to the potential abnormal data, and determines the sensing device corresponding to each potential abnormal data, so as to obtain multi-source vehicle information through the various sensing devices set in the tunnel to be controlled.

In one embodiment, there is a tunnel in a core traffic section of a city. As the traffic volume in the tunnel has continued to increase in recent years, the traffic management department has decided to use a tunnel traffic intelligent management and control system based on holographic perception to improve the safety and traffic efficiency of the tunnel.

First, the system uses its built-in geographic information module to determine that the tunnel is a tunnel to be controlled. Then, the system obtains the historical traffic information of the tunnel over the past year from the traffic management database. This information includes but is not limited to: traffic flow data, traffic accident records, violation records, average vehicle speed, congestion period, etc.

Through in-depth analysis of these historical traffic information, the system found that the tunnel is often congested during peak hours in the morning and evening, and there is a certain incidence of rear-end collisions. Therefore, the system identified two main traffic abnormal events: congestion and rear-end collisions. For congestion abnormal events, the abnormal event type is "traffic congestion"; for rear-end collisions, the abnormal event type is "traffic accident".

Next, the system further determines the types of potential abnormal data that need to be focused on and acquired based on these two types of abnormal events. For "traffic congestion", key data include vehicle speed, traffic volume and vehicle density; while for "traffic accidents", more attention should be paid to abnormal driving behaviors such as sudden braking and rapid lane changes, as well as road occupancy after the accident.

In order to obtain this multi-source vehicle information, the system analyzed the sensing devices installed in the tunnel, including high-definition cameras, radar speed guns, infrared sensors, etc., and determined the type of data that each sensing device can provide. For example, high-definition cameras can be used to capture the vehicle's driving trajectory and driving behavior, radar speed guns can provide vehicle speed data, and infrared sensors can detect vehicle density and road occupancy in the tunnel.

In one embodiment of the present application, the server performs cleaning processing on the multi-source vehicle information to remove abnormal data and duplicate data in the multi-source vehicle information, and calibrates the cleaned multi-source vehicle information; performs time synchronization processing on the cleaned and calibrated multi-source vehicle information to obtain standard multi-source vehicle information, and extracts vehicle features associated with traffic abnormalities in the standard multi-source vehicle information; and fuses the vehicle features through the Kalman filter algorithm to obtain fused vehicle multi-dimensional information. It should be noted that the vehicle multi-dimensional information in the embodiment of the present application includes at least: vehicle position, vehicle speed, acceleration and vehicle model.

The server needs to determine the measurement noise of the sensing device corresponding to the vehicle multi-dimensional information, and based on the measurement noise, determine the compensation coefficient corresponding to the vehicle multi-dimensional information, so as to compensate the vehicle multi-dimensional information according to the compensation coefficient to obtain the compensated vehicle multi-dimensional information.

The vehicle features associated with abnormal traffic events extracted by this application include vehicle position, speed, acceleration and vehicle model. It should be noted that the vehicle position in the embodiments of this application refers to the specific position of the vehicle at a certain moment. In abnormal traffic events, the vehicle position is an important basis for judging whether the vehicle is involved in accidents, violations and other behaviors. The speed of the vehicle is a key indicator for judging whether it is speeding, whether it suddenly decelerates or accelerates and other abnormal behaviors. The change in acceleration can reflect the driving state of the vehicle, such as sudden acceleration, sudden braking, etc., which are important characteristics of abnormal traffic events. In addition, the driving characteristics of different vehicle models in tunnels and the traffic problems that may be caused will also be different.

In the specific process of fusing data from multiple sensors through the Kalman filter algorithm, the first step is to establish a mathematical model for the vehicle state, including dynamic equations and observation equations. The dynamic equations describe the evolution of the vehicle state over time, such as position, speed, acceleration, etc., while the observation equations describe how to derive the vehicle state from the sensor measurements. For example, for the two-dimensional state of vehicle position and speed, the dynamic equations describe how the vehicle position and speed change over time, and the observation equations involve data obtained from GPS positioning systems or radar speed guns, and associate these data with the actual state of the vehicle.

Before starting to use the Kalman filter algorithm, the vehicle state needs to be initialized, including setting the initial estimated values of the position, speed and other states, as well as the uncertainty of these estimated values, that is, the covariance matrix. It should be noted that when the vehicle state is initialized in this application, it is set based on prior knowledge or previous measurement results. For example: when a vehicle enters the tunnel through the camera set at the entrance of the tunnel to be controlled, the initial position of the entering vehicle is set to the tunnel entrance, so as to facilitate the determination of the position of the entering vehicle and the positional relationship between the entering vehicle and the tunnel to be controlled.

Before a measurement cycle, the Kalman filter predicts the vehicle state based on the vehicle position, speed and other information input into the dynamic equation, and calculates based on the state estimate and dynamic equation of the previous time step. The prediction result is a Gaussian distribution, which represents the uncertainty of the current state estimate. When new real-time measurement data is obtained through multiple data sources, for example, the GPS positioning system provides new vehicle position information, and the radar speed meter provides new speed data, the Kalman filter compares these data with the predicted state one by one. Combine the predicted values and measured values corresponding to the obtained vehicle position and speed data, and consider the uncertainty corresponding to these data to calculate the optimal state estimate of the vehicle through the observation equation. In the measurement update step, the Kalman filter is actually fusing data, weighing the credibility of the predicted value and the measured value, and obtaining a more accurate state estimate based on their uncertainty. This process is the fusion of multiple sensor data, because the measured values may come from different sensors. After completing a measurement update, the Kalman filter can obtain a more accurate vehicle state estimate. Then, this new estimate will be used as the prediction basis for the next time step, and the cycle will be iterated to continuously improve the accuracy of the state estimate.

Specifically, the Kalman filter algorithm combines the position data provided by the GPS positioning system and the position changes of the vehicle derived from the vehicle dynamic model based on the changes in vehicle speed to output a more accurate position estimate than a single GPS positioning system. Similarly, the Kalman filter algorithm combines the data provided by the radar speed meter and the speed and acceleration predictions in the vehicle dynamic model to obtain more accurate vehicle speed and acceleration estimates.

It should be noted that the data from a single sensor may be affected by various factors, such as the sensor's own errors, environmental factors, etc. By fusing the data from multiple sensors, these errors can be averaged out, improving the accuracy and reliability of vehicle status information. If a sensor fails or the data is abnormal, the fusion algorithm can rely on the data from other sensors to continue to provide reliable vehicle status estimates, enhancing the robustness of the system. In addition, the fusion algorithm can integrate information from multiple data sources to fill in the data gaps or anomalies that may exist in a single data source, and also provide a more comprehensive description of the vehicle status, which helps to more accurately detect and respond to abnormal traffic events.

In one embodiment, in an advanced intelligent traffic management system, a series of processing steps are performed on the multi-source vehicle information collected in a tunnel to ensure the accuracy and reliability of the data, thereby improving the level of intelligence in tunnel traffic control. First, the system cleans the original multi-source vehicle information collected from various sensing devices (such as cameras, radars, infrared sensors, etc.). The purpose of this step is to remove abnormal data and duplicate data caused by sensing device failure, data transmission errors, or other reasons. For example, the system detects that the vehicle speed data at a certain point in time is abnormally high and obviously exceeds the speed limit of the tunnel. Such data is regarded as abnormal data and is cleaned. At the same time, the system also performs deduplication processing for the same data that is transmitted repeatedly.

After cleaning, the system calibrates the remaining data. The calibration process mainly adjusts the data based on the known error characteristics of the sensing device to make it closer to the true value. For example, for the data of the radar speed meter, the system calibrates it based on its inherent measurement error to ensure the accuracy of the vehicle speed data.

Next, the system performs time synchronization on the cleaned and calibrated multi-source vehicle information. Time synchronization is necessary because different sensing devices may have different sampling frequencies and data transmission delays. The system uses a high-precision time synchronization algorithm to ensure that all data can be compared and analyzed under the same time reference. After time synchronization is completed, the system obtains a set of standard multi-source vehicle information. From this information, the system further extracts vehicle features associated with abnormal traffic events. For example, for possible rear-end collisions, the system pays special attention to the vehicle's behavioral characteristics such as rapid acceleration, rapid deceleration, and rapid lane changes.

Then, the system fuses these vehicle features through the Kalman filter algorithm. The Kalman filter algorithm is an efficient recursive filter that can estimate the state of a dynamic system from a series of incomplete and noisy measurements. In this embodiment, the Kalman filter algorithm is used to fuse data from different sensing devices to obtain more accurate vehicle multi-dimensional information, including vehicle position, speed, acceleration, and vehicle model. Finally, the system also considers the impact of the measurement noise of the sensing device on the fusion result. By analyzing the measurement noise characteristics of each sensing device, the system determines the compensation coefficients corresponding to the vehicle multi-dimensional information. These compensation coefficients are used to further compensate and adjust the fused vehicle multi-dimensional information, so as to obtain more accurate compensated vehicle multi-dimensional information.

In one embodiment, the present application uses 5G communication technology to achieve communication between vehicles in the tunnel to be controlled, so that each vehicle can share information such as road conditions and accidents with each other, and provide collaborative driving suggestions for each vehicle in the tunnel to be controlled based on the information such as road conditions and accidents in the tunnel to be controlled. In addition, a human-computer interaction interface is also provided, so that managers can not only monitor the traffic conditions of the tunnel in real time, but also remotely control and intervene according to actual needs, and display traffic information in a visual way to the drivers and passengers corresponding to each vehicle in the tunnel to be controlled.

102. Based on the historical traffic information in the tunnel to be controlled and the integrated multi-dimensional information of vehicles, predict the vehicle behavior in the tunnel to be controlled.

The vehicle state model is usually based on a physical kinematic model or a dynamic model to describe the motion state of the vehicle in the tunnel. This model may include state variables such as the vehicle's position, velocity, and acceleration. For example, a simple vehicle state model can be a first-order or second-order ordinary differential equation that describes how the vehicle's position and velocity change over time. Among them, the vehicle position can use a fixed point in the tunnel as the origin, establish a one-dimensional coordinate system along the tunnel direction, and the vehicle's position can be represented by a point in this coordinate system. The vehicle speed is the change in the vehicle's position over time, such as meters per second.

Tunnel influencing factors include tunnel length, tunnel width, lighting conditions, air quality and noise level in the tunnel, traffic signs and signal lights in the tunnel, road surface flatness, road surface humidity and road surface temperature, traffic flow and vehicle type composition in the tunnel, etc., which can be determined according to actual needs and are not specifically limited in this application.

When extracting tunnel influencing factors from historical traffic information, we first collect traffic data through surveillance cameras, sensors and other equipment in the tunnel, as well as the design drawings and operation records of the tunnel. Then, we perform image processing on the collected video data to identify the vehicle type, speed and density, and obtain environmental parameters in the tunnel, such as temperature, humidity and air quality, by analyzing sensor data. We can also extract the geometric characteristics of the tunnel and traffic sign information based on the design drawings and operation records.

Through the processed data, it is possible to count the traffic flow and vehicle type composition in different time periods, calculate the average speed of vehicles traveling in the tunnel, the standard deviation of the speed and other traffic flow parameters, and extract environmental parameters such as the lighting intensity and air quality index in the tunnel. Then, the extracted features are correlated with the traffic conditions of the tunnel, and any statistical method or machine learning algorithm in the prior art is used to identify the main factors affecting tunnel traffic, that is, the tunnel influencing factors in the tunnel to be controlled. Afterwards, the extracted tunnel influencing factors and related traffic data are stored in the database for subsequent data mining and model training.

The main function of the vehicle state model is to estimate and predict the current and future state of the vehicle, including but not limited to the vehicle's position, speed, acceleration, heading and other dynamic information. With this information, the traffic management system can more effectively monitor traffic flow, predict potential safety risks, optimize the control strategy of traffic lights, and provide decision support for autonomous vehicles. The data input to the vehicle state model usually comes from a variety of sensors and information systems, such as the latitude and longitude coordinates of the vehicle provided by the GPS positioning coefficient of the vehicle traveling in the tunnel, the video frame image containing the traveling vehicle provided by the camera in the tunnel, the accelerometer and gyroscope data in the inertial measurement unit data, which can measure and calculate the acceleration, angular velocity and attitude of the vehicle, and the position and speed of the vehicle, pedestrians, obstacles, etc. detected by the radar speed gun.

The vehicle state model processes input data by fusing data from different sensors to improve the accuracy and reliability of the data. For example, by fusing GPS and IMU data, the position and attitude of the vehicle can be estimated more accurately. Use algorithms such as Kalman filtering to remove noise and outliers in sensor data to obtain smoother and more reliable data sequences. Estimate the vehicle's state variables, such as position, velocity, acceleration, etc., based on physical models (such as dynamic models) and sensor data.

In one embodiment, the vehicle state model uses the latitude and longitude information provided by GPS data to determine the absolute position of the vehicle on the earth, and the accelerometer measurements in the IMU data can be integrated to estimate the relative displacement of the vehicle, but long-term integration will lead to error accumulation. Therefore, GPS data is usually used to regularly correct these estimates. GPS data can provide an approximate speed of the vehicle, but may be affected by signal interference or sampling rate. The accelerometer data in the IMU can be integrated to calculate the speed change. However, due to the accumulation of integration errors, it is also necessary to regularly use GPS data for correction. By fusing GPS and IMU data, a more accurate and stable speed estimate can be obtained. The accelerometer in the IMU directly measures the acceleration of the vehicle. After filtering and correction, these data can provide real-time acceleration information of the vehicle. Acceleration can also be estimated indirectly by comparing the speed changes at consecutive time points.

Suppose there is a highway tunnel that is 1,000 meters long. The lighting conditions in the tunnel are good, but there is a certain amount of air pollution, such as a PM2.5 concentration of 80 micrograms per cubic meter. There is slight traffic congestion in the tunnel, the average vehicle speed is about 40 km/h, and there are multiple bends and a certain slope.

The vehicle is driving and the following data is collected: GPS data shows that the latitude and longitude of the vehicle at a certain point in time is (120.123456, 30.654321), and based on the position change at two consecutive time points, the vehicle's approximate speed is estimated to be 60km/h. The IMU accelerometer measures the vehicle's acceleration in the X, Y, and Z directions as 0.5m/ ^s2 , 0.3m/ ^s2 , and -0.1m/ ^s2, respectively (assuming the vehicle is driving on a slightly sloped road).

The vehicle state model does the following: Use GPS data to determine the exact location of the vehicle. Combine GPS and IMU data to calculate a more accurate speed of the vehicle through data fusion algorithms (such as Kalman filters). For example, if the IMU data shows that the vehicle is accelerating, but the GPS data shows that the speed has dropped slightly, the model will take both data into consideration and give a compromise speed estimate. Based on the accelerometer data of the IMU, the real-time acceleration information of the vehicle is directly obtained. At the same time, the acceleration value is indirectly verified and corrected by comparing the GPS speed data at consecutive time points. And considering the air pollution in the tunnel, the model will correct the sensor data to reduce the impact of pollutants on the accuracy of the sensor.

The observation model fuses data from different sensors to provide a more accurate estimate of the vehicle state. For example, by fusing GPS and IMU data, the position and attitude of the vehicle in a tunnel can be determined more accurately.

Taking into account the length of the tunnel, the curves and the slope, the vehicle state model adjusts its estimates of vehicle speed and acceleration to reflect the impact of these geographic features on vehicle behavior. Air pollution levels are also taken into account, as high pollutant concentrations can affect the vehicle's engine performance and the driver's vision, which can affect vehicle speed and driving behavior.

Afterwards, based on the output of the vehicle state model and the observation model, combined with the tunnel influencing factors, the system can predict the future behavior of the vehicle. For example, in a tunnel with heavy congestion and air pollution, the system may predict that the vehicle speed will drop further and remind the driver to slow down. If there is a curve ahead in the tunnel, the system will also predict that the vehicle needs to slow down and prepare to turn. Finally, the system outputs the predicted vehicle behavior to the driver or the autonomous driving system, who can make corresponding decisions based on these predictions, such as slowing down, changing lanes, or maintaining the current speed.

In one embodiment, assume that a car is driving in a tunnel with an initial speed of 60 km/h. Through the analysis of the vehicle state model and the observation model, the system finds that there is a curve ahead and the air quality in the tunnel is poor. Based on this information, the system predicts that the vehicle needs to slow down to 40 km/h to safely pass the curve and avoid accidents caused by poor visibility caused by air pollution. Therefore, the system prompts the driver to slow down.

Specifically, in one embodiment of the present application, the server determines the tunnel traffic characteristics and vehicle movement laws corresponding to the tunnel to be controlled based on the historical traffic information of the tunnel to be controlled, and extracts the tunnel influence factor corresponding to the tunnel to be controlled from the historical traffic information; establishes a vehicle state model corresponding to the tunnel to be controlled based on the tunnel traffic characteristics and the vehicle operation laws, and associates the fused vehicle multi-dimensional information with the vehicle state model to form an observation model; based on the vehicle state model and the observation model, and according to the tunnel influence factor, analyzes the vehicle multi-dimensional information to obtain the vehicle state in the tunnel to be controlled, so as to predict the vehicle behavior in the tunnel to be controlled based on the vehicle state.

In one embodiment, considering the complex traffic conditions in a tunnel, the traffic management department decided to use advanced data analysis technology to predict and manage vehicle behavior in the tunnel. First, the historical traffic information of the tunnel was deeply analyzed. This information includes traffic flow, speed distribution, accident records, weather conditions, etc. By analyzing this information, the traffic characteristics of the tunnel can be determined, such as changes in traffic flow during peak hours, the average speed of vehicles, and common driving behavior patterns. At the same time, key factors affecting tunnel traffic were also found, such as weather changes, road maintenance conditions, lighting conditions in the tunnel, etc. These factors are collectively referred to as "tunnel influencing factors."

Next, a vehicle state model was established based on the tunnel traffic characteristics and vehicle operation rules. This model can describe how the vehicle state (such as position, speed, acceleration, etc.) changes over time under specific traffic conditions. In order to verify and optimize this model, the multi-dimensional vehicle information previously fused by the Kalman filter algorithm was also associated with the model to form an observation model. This observation model not only reflects the actual motion state of the vehicle, but also takes into account the measurement errors of the perception device and the uncertainty in the data processing process.

With the vehicle state model and observation model, we used these two models to conduct an in-depth analysis of the fused vehicle multi-dimensional information, with special attention paid to the impact of tunnel influencing factors on vehicle state. For example, on rainy days, the road surface in the tunnel becomes slippery, which may lead to a reduction in vehicle speed and an increase in accident risk. By incorporating these influencing factors into the analysis, it is possible to more accurately predict vehicle behavior under different weather conditions. Finally, by combining the vehicle state model, observation model and tunnel influencing factors, the vehicle behavior in the tunnel was predicted. These predictions not only include the vehicle's future position and speed, but also involve possible driving risks and safety hazards.

103. Determine the vehicle's motion trajectory based on the vehicle's multi-dimensional information, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle's driving data and vehicle motion trajectory in the vehicle's multi-dimensional information, and predict the traffic safety risk points in the tunnel to be controlled based on the vehicle's behavior and traffic situation.

Specifically, the server determines the sampling position and sampling time corresponding to the vehicle multi-dimensional information to generate a data sampling trajectory corresponding to the vehicle multi-dimensional information; based on the data sampling trajectory, the vehicle position and speed in the vehicle multi-dimensional information are fitted to determine the vehicle motion trajectory of each vehicle in the tunnel to be controlled; based on the vehicle multi-dimensional information and the vehicle motion trajectory, the traffic flow parameters in the tunnel to be controlled are calculated to perceive the traffic situation in the tunnel to be controlled in real time based on the traffic flow parameters and vehicle behavior. It should be noted that the traffic flow parameters in the embodiment of the present application include vehicle flow, average speed, vehicle density, and road occupancy.

In one embodiment, in a tunnel traffic management project, in order to more accurately grasp the traffic situation in the tunnel, the management department adopted advanced trajectory analysis and traffic flow parameter calculation methods. First, the system determines the sampling position and sampling time of each data point in the fused vehicle multi-dimensional information. This information is obtained by sampling the vehicle at different time points through the sensing device in the tunnel, so each data point corresponds to a specific spatial position and timestamp. By integrating these data points, the system generates the data sampling trajectory of each vehicle in the tunnel.

Next, the system used these data sampling trajectories to fit the position and speed of the vehicle. During the fitting process, the system used high-precision mathematical models and algorithms to ensure that the generated vehicle motion trajectory can reflect the actual movement of the vehicle as realistically as possible. In this way, the system successfully determined the vehicle motion trajectory of each vehicle in the tunnel.

With these vehicle movement trajectories, the system begins to calculate the traffic flow parameters in the tunnel. These parameters include vehicle volume, average vehicle speed, vehicle density, and road occupancy, which are important bases for evaluating the traffic situation in the tunnel. The system accurately calculates these traffic flow parameters through in-depth analysis of vehicle multi-dimensional information and vehicle movement trajectories. For example, the vehicle volume is obtained by counting the number of vehicles passing through the tunnel in a specific time period; the average vehicle speed is obtained by calculating the average of all vehicle speeds; the vehicle density is calculated based on the number of vehicles in the tunnel and the total area of the tunnel; and the road occupancy is determined by analyzing the vehicle movement trajectory and the spatial layout of the tunnel. Finally, based on these traffic flow parameters and previously predicted vehicle behavior, the system perceives the traffic situation in the tunnel in real time.

In one embodiment of the present application, the server inputs the vehicle behavior into the observation model to determine the vehicle behavior and reward value corresponding to each vehicle in the tunnel to be controlled at the next moment; determines the descent gradient corresponding to each vehicle in the tunnel to be controlled at the next moment based on the traffic situation; inputs the vehicle behavior, reward value and descent gradient corresponding to each vehicle at the next moment into the pre-trained safety identification model to determine the risky vehicles with potential risks in the tunnel to be controlled, and the vehicle positions corresponding to the risky vehicles, and determines the traffic safety risk points in the tunnel to be controlled based on the vehicle positions.

In one embodiment, assume that there is a highway tunnel in which multiple sensors and cameras are installed to monitor the traffic conditions in the tunnel in real time. The observation model receives real-time data from the sensors and cameras in the tunnel. By analyzing these data, the observation model can predict the possible vehicle behavior (such as acceleration, deceleration, lane change, etc.) and the corresponding reward value of each vehicle in the tunnel at the next moment. The reward value is a quantitative indicator used to evaluate the rationality and safety of vehicle behavior.

Based on the vehicle behavior and reward values output by the observation model, the system further analyzes the traffic situation of the entire tunnel. By comparing the relative speed, position, and driving direction of different vehicles, the system calculates the possible descent gradient of each vehicle at the next moment. The calculated descent gradient reflects the trend of increased potential risks caused by vehicle behavior.

The safety recognition model is a trained deep learning model obtained through training based on historical sample data. The trained safety recognition model receives vehicle behavior predictions, reward values, and descent gradients from the observation model as input. In this way, vehicles with potential risks can be directly identified through the trained safety recognition model, and the identification of vehicles identified as risky vehicles and their specific locations can be output.

Assume that at a certain moment, the observation model predicts that a car (vehicle A) in the tunnel is rapidly approaching another car (vehicle B) in front of it, while vehicle B is slowing down to enter the parking area on the right. The observation model calculates that the reward value of vehicle A is low because its high-speed approach to the front car may lead to a rear-end collision. At the same time, the traffic situation analysis shows that the descent gradient of vehicle A is steep, which means that if no action is taken, the risk will increase rapidly.

This information was input into the safety recognition model, which determined that vehicle A was a potential risk vehicle based on historical data and algorithms. The model further outputted the specific location of vehicle A, and the tunnel management system immediately issued an alarm and notified the driver to slow down through variable message signs or broadcasts, thus avoiding a possible rear-end collision.

In one embodiment, in the traffic safety management of a certain tunnel, in order to predict and identify potential risk vehicles and traffic safety risk points, the management department adopts an intelligent analysis method based on an observation model and a safety identification model.

First, the management department inputs the previously predicted vehicle behavior into the observation model. This observation model is built based on historical data and machine learning algorithms. It can predict the vehicle behavior at the next moment based on the current vehicle behavior and give a reward value. The reward value reflects the safety and efficiency of the vehicle behavior. For example, maintaining a stable driving speed and appropriate distance between vehicles will receive a higher reward value, while frequent lane changes or speeding may result in a lower reward value.

Next, the management department determines the corresponding descent gradient for each vehicle in the tunnel at the next moment based on the real-time traffic situation. The descent gradient reflects the degree of deterioration of the traffic situation. For example, in congested or accident-prone areas, the descent gradient will be relatively high. This parameter helps identify vehicles that may face higher risks due to changes in traffic conditions.

The management department then inputs the predicted vehicle behavior, reward value, and descent gradient into a pre-trained safety identification model. This safety identification model is built based on a deep learning algorithm, which can identify risky vehicles with potential risks from the input data. The model analyzes the comprehensive information of vehicle behavior, reward value, and descent gradient to determine which vehicles may face safety risks in the future.

Finally, through the output of the safety identification model, the management department successfully identified the risky vehicles with potential risks in the tunnel and their specific locations. Based on this information, the management department can further identify the traffic safety risk points in the tunnel, which may be accident-prone areas, areas with severe congestion, or areas with abnormal driving behavior. In addition, targeted measures can be taken quickly to reduce safety risks. For example, adding monitoring equipment at risk points, strengthening patrols, setting up warning signs, or conducting traffic diversion. These measures help to discover and deal with potential safety hazards in advance, thereby ensuring the traffic safety and smooth operation of the tunnel.

104. Based on the traffic safety risk points, the tunnel control strategy is adjusted to achieve intelligent control of tunnel traffic according to the optimized tunnel control strategy.

Specifically, the server determines the corresponding area to be controlled in the tunnel to be controlled according to the traffic safety risk point, and obtains the multi-dimensional information of vehicles and the actual situation of the road section in the area to be controlled. It should be noted that the actual situation of the road section in the embodiment of the present application includes the existence of obstacles, traffic accidents, and the intrusion of pedestrians and non-motor vehicles.

The server needs to determine the impact coefficient of the actual road situation on each vehicle in the controlled area, combine the impact coefficient, and determine the corresponding vehicle behavior based on the multi-dimensional information of the vehicles in the controlled area; determine the risk event type based on the vehicle behavior in the controlled area to generate tunnel control instructions corresponding to the risk event type, and adjust the traffic flow in the controlled area based on the tunnel control instructions to realize intelligent control of tunnel traffic.

In one embodiment, during the traffic safety management process of a certain tunnel, the management department has taken a series of measures to achieve intelligent control of specific areas. First, based on the previously determined traffic safety risk points, the management department has identified the areas to be controlled in the tunnel. These areas are usually places where accidents are frequent or traffic is congested. Subsequently, through the sensing device in the tunnel, multi-dimensional information of vehicles in these areas to be controlled is obtained, including vehicle position, speed, acceleration, etc., and the actual situation of the road section is obtained at the same time, such as whether there are obstacles, whether a traffic accident has occurred, whether pedestrians or non-motor vehicles have intruded, etc.

Next, the impact of the actual conditions of the road section on the vehicles in the area to be controlled was evaluated. For example, if a traffic accident occurs in a certain area, the impact coefficient of the accident on the surrounding vehicles will increase accordingly. By comprehensively considering the actual conditions of various road sections and their impact coefficients, the management department can more accurately predict and judge the behavior of vehicles. Then, combining the impact coefficient and the multi-dimensional information of the vehicle, the specific behavior of each vehicle in the area to be controlled was determined, such as deceleration, lane change, parking, etc. These vehicle behavior predictions provide an important basis for the subsequent generation of control instructions.

Finally, based on the predicted vehicle behavior, possible risk event types are determined, such as rear-end collisions and side-slip accidents, and corresponding tunnel control instructions are generated for these risk events. These instructions may include speed limits, lane change prohibitions, turning on warning lights, etc., aiming to reduce accident risks by regulating traffic flow and ensure the safety and smoothness of tunnel traffic.

In one embodiment of the present application, after the tunnel control strategy is adjusted based on the traffic safety risk points to achieve intelligent control of tunnel traffic according to the optimized tunnel control strategy, the server obtains the traffic feedback data in the area to be controlled in real time through the sensing device of the area to be controlled, and determines the execution effect corresponding to the tunnel control instruction based on the traffic feedback data; determines the loss function corresponding to the tunnel control instruction based on the deviation between the execution effect and the expected control effect of the tunnel, so as to optimize the tunnel control instruction according to the loss function.

In one embodiment, in a tunnel's intelligent traffic control system, in order to continuously optimize control instructions and improve the management effect of tunnel traffic, the management department adopts an optimization method based on real-time traffic feedback data. First, through the sensing devices in the area to be controlled, such as cameras, sensors, etc., the system obtains the traffic feedback data in the area to be controlled in real time. These data include vehicle speed, traffic flow, vehicle queue length, traffic accident occurrence, etc., which can intuitively reflect the execution effect of tunnel control instructions.

Next, based on these traffic feedback data, the actual implementation effect of the tunnel control instructions was evaluated. For example, if the vehicle speed drops significantly after the speed limit instruction is issued, then it can be considered that the instruction has been effectively implemented. At the same time, the expected control effect of the tunnel is also set, that is, the traffic state expected to be achieved through the control instruction. Then, the actual implementation effect is compared with the expected control effect, and the deviation between the two is calculated. This deviation reflects the shortcomings of the control instruction in the execution process and also provides a direction for subsequent optimization.

In order to quantify this deviation and find the direction of optimization, the loss function corresponding to the tunnel control instructions was also determined. This loss function comprehensively considers multiple aspects such as vehicle speed, traffic flow, and safety. By minimizing the loss function, a better control instruction can be found. Finally, according to the optimization results of the loss function, the tunnel control instructions were adjusted and optimized. For example, if it is found that the speed limit instruction is too strict and causes traffic congestion, the speed limit value can be appropriately increased; if it is found that traffic accidents occur frequently in a certain area, the monitoring and warning measures in that area can be increased.

The above is an embodiment of the method proposed in this application. Based on the same inventive concept, the embodiment of this application also provides a tunnel traffic intelligent control device based on holographic perception, and its structure is shown in FIG2 .

FIG2 is a schematic diagram of the internal structure of a tunnel traffic intelligent control device based on holographic perception provided in an embodiment of the present application. As shown in FIG2 , the device includes:

at least one processor;

and, a memory communicatively coupled to the at least one processor;

The memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor to enable the at least one processor to:

Obtain multi-source vehicle information in the tunnel to be controlled through multiple data sources, and fuse the multi-source vehicle information corresponding to each vehicle based on the Kalman filter algorithm;

Based on the historical traffic information and the integrated multi-dimensional vehicle information in the tunnel to be controlled, the vehicle behavior in the tunnel to be controlled is predicted;

Determine the vehicle movement trajectory according to the vehicle multi-dimensional information, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle driving data and vehicle movement trajectory in the vehicle multi-dimensional information, and predict the traffic safety risk points in the tunnel to be controlled according to the vehicle behavior and traffic situation;

Based on the traffic safety risk points, the tunnel control strategy is adjusted to achieve intelligent control of tunnel traffic according to the optimized tunnel control strategy.

The present application also provides a non-volatile computer storage medium storing computer executable instructions, which can:

Obtain multi-source vehicle information in the tunnel to be controlled through multiple data sources, and fuse the multi-source vehicle information corresponding to each vehicle based on the Kalman filter algorithm;

Based on the historical traffic information and the integrated multi-dimensional vehicle information in the tunnel to be controlled, the vehicle behavior in the tunnel to be controlled is predicted;

Determine the vehicle movement trajectory according to the vehicle multi-dimensional information, so as to perceive the traffic situation in the tunnel to be controlled in real time based on the vehicle driving data and vehicle movement trajectory in the vehicle multi-dimensional information, and predict the traffic safety risk points in the tunnel to be controlled according to the vehicle behavior and traffic situation;

Based on the traffic safety risk points, the tunnel control strategy is adjusted to achieve intelligent control of tunnel traffic according to the optimized tunnel control strategy.

Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

The above describes specific embodiments of the present application. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

The present application is described with reference to the flowchart and/or block diagram of the method, device (system) and computer program product according to the embodiment of the present application. It should be understood that each process and/or box in the flowchart and/or block diagram, and the combination of the process and/or box in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for realizing the function specified in one process or multiple processes in the flowchart and/or one box or multiple boxes in the block diagram.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and/or one or more boxes in the block diagram.

In a typical configuration, a computing device includes one or more processors (CPU), input/output interfaces, network interfaces, and memory.

The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and/or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims (10)

1. The intelligent tunnel traffic control method based on holographic perception is characterized by comprising the following steps of:

The method comprises the steps of obtaining multi-source vehicle information in a tunnel to be controlled through a plurality of data sources, and fusing the multi-source vehicle information corresponding to each vehicle based on a Kalman filtering algorithm;

Predicting the vehicle behavior in the tunnel to be controlled based on the historical traffic information in the tunnel to be controlled and the fused vehicle multidimensional information;

Determining a vehicle motion trail according to the vehicle multidimensional information, sensing traffic situation in the tunnel to be controlled in real time based on vehicle running data in the vehicle multidimensional information and the vehicle motion trail, and predicting traffic safety risk points in the tunnel to be controlled according to the vehicle behavior and the traffic situation;

And adjusting the tunnel control strategy based on the traffic safety risk points so as to realize intelligent tunnel traffic control according to the optimized tunnel control strategy.

2. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein the fusion of the multi-source vehicle information corresponding to each vehicle based on the kalman filtering algorithm specifically comprises:

Cleaning the multi-source vehicle information to remove abnormal data and repeated data in the multi-source vehicle information, and calibrating the cleaned multi-source vehicle information;

performing time synchronization processing on the cleaned and calibrated multi-source vehicle information to obtain standard multi-source vehicle information, and extracting vehicle characteristics associated with traffic abnormal events from the standard multi-source vehicle information;

fusing the vehicle characteristics through a Kalman filtering algorithm to obtain fused vehicle multidimensional information; wherein the vehicle multidimensional information comprises at least: vehicle position, vehicle speed, acceleration and vehicle model;

And determining measurement noise of the sensing device corresponding to the vehicle multidimensional information, and determining a compensation coefficient corresponding to the vehicle multidimensional information based on the measurement noise so as to compensate the vehicle multidimensional information according to the compensation coefficient, thereby obtaining compensated vehicle multidimensional information.

3. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein the predicting the vehicle behavior in the tunnel to be controlled based on the historical traffic information in the tunnel to be controlled and the merged vehicle multidimensional information specifically comprises:

according to the historical traffic information of the tunnel to be managed, determining the tunnel traffic characteristics and the vehicle movement rules corresponding to the tunnel to be managed, and extracting tunnel influence factors corresponding to the tunnel to be managed from the historical traffic information;

According to the tunnel traffic characteristics and the vehicle running rules, building a vehicle state model corresponding to the tunnel to be controlled, and associating the fused vehicle multidimensional information with the vehicle state model to form an observation model;

And analyzing the vehicle multidimensional information based on the vehicle state model and the observation model according to the tunnel influence factors to obtain the vehicle state in the tunnel to be managed so as to predict the vehicle behavior in the tunnel to be managed based on the vehicle state.

4. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein the determining a vehicle motion track according to the vehicle multidimensional information so as to sense traffic situation in the tunnel to be controlled in real time based on vehicle running data in the vehicle multidimensional information and the vehicle motion track specifically comprises:

determining sampling positions and sampling times corresponding to the vehicle multidimensional information to generate a data sampling track corresponding to the vehicle multidimensional information;

fitting the vehicle position and the vehicle speed in the vehicle multidimensional information based on the data sampling track to determine the vehicle motion track of each vehicle in the tunnel to be controlled;

Calculating traffic flow parameters in the tunnel to be controlled according to the vehicle multidimensional information and the vehicle motion trail so as to sense traffic situations in the tunnel to be controlled in real time based on the traffic flow parameters and the vehicle behaviors; wherein the traffic flow parameters include vehicle flow, average vehicle speed, vehicle density, and road occupancy.

5. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein predicting the traffic safety risk point in the tunnel to be controlled according to the vehicle behavior and the traffic situation specifically comprises:

Inputting the vehicle behaviors into an observation model to determine vehicle behaviors and rewarding values corresponding to the vehicles in the tunnel to be managed at the next moment;

Determining a descent gradient corresponding to each vehicle in the tunnel to be managed at the next moment according to the traffic situation;

And inputting the vehicle behaviors, the rewarding values and the descending gradients corresponding to the vehicles at the next moment into a pre-trained safety recognition model to determine a risk vehicle with potential risks in the tunnel to be managed and a vehicle position corresponding to the risk vehicle, and determining traffic safety risk points in the tunnel to be managed according to the vehicle position.

6. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein the adjusting the tunnel control policy based on the traffic safety risk point specifically comprises:

Determining a corresponding region to be controlled in the tunnel to be controlled according to the traffic safety risk points, and acquiring vehicle multidimensional information and road section live conditions of the region to be controlled; the road section live condition comprises the existence of an obstacle, the occurrence of a traffic accident and the intrusion of pedestrians and non-motor vehicles;

Determining influence coefficients of the road section live condition on each vehicle in the area to be controlled so as to combine the influence coefficients, and determining corresponding vehicle behaviors according to the vehicle multidimensional information in the area to be controlled;

And determining a risk event type according to the vehicle behavior of the area to be controlled so as to generate a tunnel control instruction corresponding to the risk event type, and adjusting the traffic flow of the area to be controlled according to the tunnel control instruction so as to realize intelligent control of tunnel traffic.

7. The intelligent tunnel traffic control method based on holographic sensing according to claim 6, wherein the adjusting the tunnel control policy based on the traffic safety risk point is performed to realize intelligent tunnel traffic control according to the optimized tunnel control policy, and the method further comprises:

Acquiring traffic feedback data in the area to be controlled in real time through a sensing device of the area to be controlled, and determining an execution effect corresponding to the tunnel control instruction according to the traffic feedback data;

And determining a loss function corresponding to the tunnel control instruction according to the deviation between the execution effect and the tunnel expected control effect, so as to optimize the tunnel control instruction according to the loss function.

8. The intelligent tunnel traffic control method based on holographic sensing according to claim 1, wherein the acquiring the multi-source vehicle information in the tunnel to be controlled through the plurality of data sources specifically comprises:

Determining a tunnel to be managed and controlled, acquiring historical traffic information corresponding to the tunnel to be managed and controlling, and determining traffic abnormal events in the tunnel to be managed and controlling and abnormal event types corresponding to the traffic abnormal events according to the historical traffic information;

And determining the data type corresponding to the potential abnormal data based on the abnormal event type, and determining a sensing device corresponding to each potential abnormal data so as to acquire multi-source vehicle information through a plurality of sensing devices arranged in the tunnel to be managed.

9. Tunnel traffic intelligent management and control equipment based on holographic perception, characterized in that the equipment includes:

At least one processor;

And a memory communicatively coupled to the at least one processor;

Wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a holographic-aware-based tunnel traffic intelligent management method according to any of claims 1-8.

10. A non-transitory computer storage medium storing computer executable instructions which, when executed, implement a holographic-awareness based intelligent tunnel traffic management and control method according to any of claims 1-8.