CN118250167B - Network resource real-time scheduling method based on multi-objective optimization algorithm - Google Patents
Network resource real-time scheduling method based on multi-objective optimization algorithm Download PDFInfo
- Publication number
- CN118250167B CN118250167B CN202410675779.0A CN202410675779A CN118250167B CN 118250167 B CN118250167 B CN 118250167B CN 202410675779 A CN202410675779 A CN 202410675779A CN 118250167 B CN118250167 B CN 118250167B
- Authority
- CN
- China
- Prior art keywords
- optimization algorithm
- strategy
- module
- scheduling
- network resource
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0876—Aspects of the degree of configuration automation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0894—Policy-based network configuration management
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0895—Configuration of virtualised networks or elements, e.g. virtualised network function or OpenFlow elements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
- H04L67/125—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks involving control of end-device applications over a network
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Physiology (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Genetics & Genomics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Automation & Control Theory (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a network resource real-time scheduling method based on a multi-objective optimization algorithm, which comprises the following steps: the physical world collects real-time state data of the real monitoring system platform and flows the collected information into the CPS event model, the twin database and the database of the unified management platform respectively; the CPS event model takes the preprocessed data as input, transmits the preprocessed data to the virtual world, builds a three-dimensional simulation scene, and outputs an optimal scheduling strategy and configuration parameters required by a unified management platform according to the current state in the simulation environment; before executing the scheduling policy, invoking rules configured by the unified management platform, and determining that the current policy is executable; and after the optimal strategy is executed in the simulation environment, feeding back the network resource scheduling strategy to the physical world. According to the invention, the stability and response speed of video monitoring can be improved, and the coping capability of emergency is enhanced, so that the safety management level is improved and the potential risk is reduced.
Description
Technical Field
The invention relates to the technical field of network resource allocation, in particular to a network resource real-time scheduling method based on a multi-objective optimization algorithm.
Background
In the production and management of hydropower plants, real-time video monitoring is a key component to ensure safety and efficiency. The existing video monitoring system often depends on a static network resource allocation strategy, which easily causes unstable video monitoring quality of a key area under the conditions of limited resources and multi-task processing requirements, and particularly, when bandwidth is limited or important events happen in multiple areas at the same time, the traditional static allocation method cannot flexibly cope with the situation.
In addition, the existing system cannot effectively utilize the multi-objective optimization algorithm, and it is difficult to balance the resource use of the whole network and the minimum requirement of non-critical area monitoring on the premise of ensuring the critical area monitoring. In emergency situations, this deficiency may lead to delayed response of the monitoring system and even failure of the monitoring, thereby affecting the normal operation and safety of the hydropower plant.
Disclosure of Invention
In view of the above, the invention provides a network resource real-time scheduling method based on a multi-objective optimization algorithm, which aims to improve the video data transmission quality and efficiency in a monitoring system by dynamically optimizing the distribution of network bandwidth under the condition of limited network resources.
The invention discloses a network resource real-time scheduling method based on a multi-objective optimization algorithm, which comprises the following steps: the physical world collects real-time state data of the real monitoring system platform and flows the collected information into the CPS event model, the twin database and the database of the unified management platform respectively; the real-time state data comprise monitoring picture quality, network bandwidth occupation condition and network resource allocation condition of different monitoring areas;
The CPS event model takes the preprocessed data as input, transmits the preprocessed data to a virtual world, builds a three-dimensional simulation scene, and outputs an optimal scheduling strategy and configuration parameters required by a unified management platform according to the current state in the simulation environment so that an optimization target in a multi-target optimization algorithm is optimal; before executing the scheduling strategy, calling a rule configured by a unified management platform, determining that the current strategy is executable, and if the current strategy conflicts with the rule configuration, adjusting the scheduling strategy configuration, and outputting an optimal strategy again based on a multi-objective optimization algorithm; and after the optimal strategy is executed in the simulation environment, feeding back the network resource scheduling strategy to the physical world.
The system shown in fig. 1 distributes network bandwidth in real time according to the importance of different monitoring areas, ensures that the critical areas can obtain enough network resources to perform high-quality video monitoring, and simultaneously maintains the stability and efficiency of non-critical area monitoring. The core point to be protected is the method for dynamically adjusting network resource allocation according to real-time monitoring requirements and the application of the method in the aspects of ensuring video quality of key monitoring areas and reducing the waste of overall network resources. This includes the use of multi-objective optimization algorithms to evaluate and optimize network resource allocation, while combining digital twinning techniques for prediction and simulation, and integrating decision support systems to improve operational efficiency and response speed.
Further, the CPS event model is responsible for preprocessing the original acquired data and classifying fine granularity, wherein the data preprocessing comprises normalization and denoising of the data, and the fine granularity classification can divide the data into physical object events, control unit events, sensor events and execution equipment events; the physical object event refers to an action event of an entity object in the monitoring video; the control unit event refers to a real-time state event of a device in the physical world, and the sensor event refers to action event and state information executed by a sensor in the physical world; the executing device event refers to whether the device executing the control policy has executed the network bandwidth adjustment policy, and execution status information.
Further, a virtual simulation system in the virtual world constructs a three-dimensional simulation scene using data received from the CPS event model; the virtual world has four functions of connection, perception, decision and control, wherein the connection means mapping the real environment state of the physical world and feeding back the optimization strategy simulated in the virtual world to the physical world; the sensing means that real-time sensing data of the physical world is transmitted to the virtual world for three-dimensional reconstruction and simulation; the decision refers to a decision scheduling scheme obtained by the virtual world based on the CPS event model and the multi-objective optimization algorithm, and the decision scheduling scheme is fed back to the physical world after the virtual world is verified to be free of errors; control refers to the mapping of the virtual world itself to the physical world.
Further, the virtual world integrates four functional sub-modules: the system comprises a strategy scheduling module, a three-dimensional reconstruction module, an optimization algorithm configuration module and a control task configuration module;
the input of the strategy scheduling module is the current monitoring quality state information of the virtual world, the output is the scheduling strategy, the scheduling strategy is issued to the control task, and the control task configuration module issues the control instruction to the specific equipment for execution; the three-dimensional reconstruction configuration module is the functional configuration of the digital twin when the virtual world rendering and construction are carried out, the input of the three-dimensional reconstruction configuration module is the data which is output by the CPS model and used for constructing the twin model, the output is the configuration related information of the constructed twin body, and the three-dimensional reconstruction configuration module is completed based on the configurations when the three-dimensional simulation model is constructed; the optimization algorithm configuration is to select an optimization target and an optimization algorithm, and meanwhile, an optimization algorithm in a multi-target optimization algorithm module based on a twin database is designated through an optimization algorithm configuration module, and the control task configuration is specific parameter configuration of a control task in a digital twin world, wherein the specific parameter configuration comprises a task execution sequence and execution time limitation; the input of the control task configuration module is a strategy of strategy scheduling output, the output is a specific execution task, and the specific execution task is issued to the entity object of the virtual world for action execution.
Further, the multi-objective optimization algorithm module mainly comprises an optimization objective module and an optimization algorithm module, wherein the optimization algorithm module introduces a genetic algorithm and a chebyshev decision, and the optimization algorithm module is responsible for synthesizing three optimization objectives to obtain an optimal network resource allocation strategy and related scheduling parameters;
When the network bandwidths of different monitoring areas are required to be set, so that three optimization targets are comprehensively optimal, the network bandwidths at the moment are parameters, the optimal bandwidth parameters are output through iteration of an optimization algorithm, the bandwidth parameters are issued and bandwidth redistribution is executed, and then whether the current monitoring quality is optimized is evaluated.
Further, a multi-objective optimization algorithm in the multi-objective optimization algorithm module is combined with a chebyshev decision process to obtain an optimal network resource scheduling strategy; optimizing an objective function of three optimization objectives using a multi-objective optimization algorithm and chebyshev decision process, comprising:
step 101: initializing a population: randomly generating a series of initial solutions as a population, wherein each solution represents a network resource scheduling scheme;
step 102: objective function evaluation: for each solution in the population, three objective function values corresponding to the solution are calculated 、、；
Step 103: non-dominant ordering: non-dominant sorting is carried out on the population, and Pareto front grades of all solutions are determined;
step 104: and (3) calculating the crowding degree: in each Pareto front, calculating the congestion degree of the solutions, and selecting solutions with low congestion degree so as to keep the diversity of the population;
step 105: selection, crossover and mutation: applying the selection, crossover and mutation operations of the genetic algorithm to generate a new generation solution; in the selection process, individuals with non-dominant solutions and low crowding degrees are more prone to be selected;
Step 106: elite strategy: merging the non-dominant solution of the previous generation and the solution generated at present, and carrying out non-dominant sorting and congestion degree calculation again to ensure that elite solution is not lost;
Step 107: chebyshev decision process: for a Pareto optimal solution set found by an NSGA-III algorithm, performing weight distribution and selection on the Pareto optimal solution by using a Chebyshev method; setting a weight for each objective function, calculating the weighted objective function values to form Chebyshev functions, and selecting a solution which minimizes the Chebyshev functions as a final decision scheme;
step 108: the iterative process: repeating steps 103 to 107 until a termination condition is met; the termination condition includes reaching a maximum number of iterations or convergence of objective function values;
step 109: and (3) outputting results: the solution that minimizes the chebyshev function is output as the optimal network resource scheduling policy.
Further, the step 102 includes:
optimizing by allocating more bandwidth resources to critical areas To maximize key region video quality; setting higher weight for critical areaGiving higher priority to the key region in the multi-target optimization algorithm, and ensuring that the resource allocation of the key region is not lower than a set threshold value;
When resources are tensed, the key areas are preferentially considered, the average allocation of the resources is realized by a method of minimizing variance, and the requirements of the key areas and the allocation of the resources of other areas are balanced by setting weight coefficients;
Setting a quality threshold, introducing a punishment function in the optimization process of the multi-objective optimization algorithm, and when the video quality of the non-critical area is lower than the quality threshold, negatively affecting the overall optimization result to force the multi-objective optimization algorithm to allocate enough resources to meet the video quality requirement of the non-critical area.
Wherein,For the maximization of the video quality of the key areas, K represents the set of all key areas,Is the weight of the critical area i,Is the video quality of region i under network resource allocation x;
Wherein, For the balance of overall network resource usage, N is the total number of areas monitored,Is the resource allocated to region i, T is the total network resource available;
Wherein, Is the minimum guarantee level of video quality for non-critical areas,Representing a set of all non-critical areas,Is the video quality of region j at network resource allocation x.
Further, the method comprises the steps of,Resource allocation aimed at minimizing individual areasThe absolute difference from the average value, but taking into account that the monitoring of the critical area requires more resources, a weight coefficient can be introducedResource allocation for each region iWeighting is carried out; weight coefficientReflecting the importance of region i; setting a higher weight coefficient for the key area and setting a lower weight coefficient for the non-key area;
Involving weighting coefficients The modification is as follows:
When region i Lower than average resource allocationIf the region i is a key region, the multi-objective optimization algorithm tends to allocate more resources to the region i; otherwise, if the region i is a non-critical region, the multi-objective optimization algorithm allows the resources allocated to the region i to be lower than the average resources.
Further, the expression of the chebyshev function is:
Wherein, Is an objective functionIs used for the weight of the (c),Is an objective functionIdeal values in the Pareto optimal solution set.
Further, the whole system is regulated and controlled through a unified management platform, wherein the unified management platform comprises a database realization module, an information authority management module, a system optimization module, a visualization module, a rule triggering module and a safe operation and maintenance module; wherein the rule triggering function is related to digital twin decision; and when the twin simulation environment executes the network resource scheduling strategy output by the multi-objective optimization algorithm module, the rule triggering module is called, and then whether the scheduling strategy is executed is determined.
Due to the adoption of the technical scheme, the invention has the following advantages:
1. The invention has the practical value of improving the safety monitoring efficiency, can effectively reduce the occurrence probability of safety accidents by ensuring that the key area always obtains enough monitoring resources, and enhances the timeliness and the accuracy of safety management of a hydropower plant. In support of decision making, the integrated DSS is able to provide data analysis, prediction and visualization tools for operators, helping them make more efficient decisions, especially in emergency situations. The real-time adjustment capability of the system allows it to automatically reallocate resources as network conditions change, ensuring the stability and responsiveness of the monitoring system. The human error is reduced, the dependence on manual intervention is reduced by automatic network resource scheduling, and the error caused by improper manual operation is reduced. The network resource utilization is optimized, the multi-objective optimization algorithm can ensure that the network resources are reasonably distributed to all monitoring areas, so that resource waste is avoided, and the overall utilization efficiency of the network resources is improved.
2. The invention has economic value, reduces potential loss, and helps to prevent safety accidents which can cause great economic loss by improving monitoring quality and real-time performance. The operation cost is reduced, and the automatic and optimized resource scheduling reduces the requirement for additional network bandwidth and manpower, thereby reducing the operation cost. The service life of the equipment is prolonged, the service life of monitoring equipment and other key infrastructures can be prolonged through effective monitoring and maintenance, and maintenance and replacement cost is reduced.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments described in the embodiments of the present invention, and other drawings may be obtained according to these drawings for those skilled in the art.
FIG. 1 is a schematic diagram of a network resource real-time scheduling system based on a multi-objective optimization algorithm according to an embodiment of the present invention;
FIG. 2 is a flow chart of a multi-objective optimization algorithm according to an embodiment of the present invention;
fig. 3 is a schematic flow chart of chebyshev decision in an embodiment of the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings and examples, wherein the examples are shown only in a partial, but not in all embodiments of the invention. All other embodiments obtained by those skilled in the art are intended to fall within the scope of the embodiments of the present invention.
Referring to fig. 1, the invention provides an embodiment of a network resource real-time scheduling method based on a multi-objective optimization algorithm, which relates to a real physics, a CPS event model, a virtual world, a unified management platform and a multi-objective optimization algorithm module of a hydropower plant, and comprises:
Physical world: and collecting real-time state data of a real-time monitoring system platform of the hydropower plant, such as monitoring picture quality, network bandwidth occupation condition, network resource allocation condition and the like of different monitoring areas. These data will be used to construct a twin simulation system and train an algorithm model.
The information data collected in the physical world flows into the CPS event model, the twin database (here, the database can be written) and the database of the unified management platform, respectively. The CPS event model is responsible for preprocessing original acquired data and fine-granularity classification, wherein the data preprocessing comprises normalization, denoising and the like, and the fine-granularity classification can further divide the data into physical object events, control unit events, sensor events and execution equipment events. Physical object events refer to action events of physical objects in a monitoring video, such as departure of a person in a factory building, and the like; the control unit event refers to real-time state events such as a switch, a gateway and the like in the invention, for example, a router increases the bandwidth of a certain link, and the switch performs link aggregation operation and the like; the sensor event refers to action event and state information executed by sensors such as an IP camera and a network flow analyzer in the physical world; the executing device event refers in the present invention to a device, such as a router, a switch, etc., that executes a control policy, whether it has executed a network bandwidth adjustment policy, and execution status information.
The CPS event model takes the preprocessed data as input and transmits the preprocessed data to the virtual world, and the virtual simulation system builds a three-dimensional simulation scene by utilizing the data based on the software such as Unity3D, solidWorks and the like. The digital world has four functions of connection, perception, decision and control, the connection refers to mapping the real environment state of the physical world, and the simulation optimization strategy in the digital world is fed back to the physical world; the sensing refers to real-time sensing data of the physical world and is transmitted to the digital world for three-dimensional reconstruction and simulation; the decision index number world is based on a CPS event model and a decision scheduling scheme obtained by a multi-objective optimization algorithm, and the decision index number world is fed back to the physical world after the digital world is verified to be free of errors; the control index refers to a map of the physical world itself, so that a control policy in the physical world may reflect the physical world, for example, after the digital world executes a network scheduling control policy, observe the monitoring quality of the digital world, and if the monitoring performance is improved, it means that the control policy is valid and may be fed back to the physical world. Meanwhile, four functional sub-modules are integrated in the digital simulation world: policy scheduling, three-dimensional reconstruction, optimization algorithm configuration and control task configuration. Wherein, 1) the strategy scheduling refers to a network scheduling scheme configured for a digital twin world in the invention, such as priority scheduling strategy, weighted polling scheduling strategy and the like according to different key areas. The input of the strategy scheduling is the current monitoring quality state information of the digital world, the output is the scheduling strategy, the scheduling strategy is issued to the control task, and the control task configuration module issues the control instruction to the specific equipment for execution. 2) The three-dimensional reconstruction configuration is a functional configuration of digital twin when the digital world rendering and construction is performed, such as physical attribute simulation, geometric model selection, whether dynamic simulation is started or not, and the like. The input of the three-dimensional reconstruction configuration is the data which is output by the CPS model and used for constructing the twin model, the output is the configuration related information of the constructed twin body, and the construction of the three-dimensional simulation model is completed based on the configuration. 3) The optimization algorithm configuration refers to relevant configuration and selection of optimization targets, optimization algorithm selection and the like, such as three optimization targets selected by the invention: a) Maximizing video quality in key areas; b) And c) minimum guarantee of video quality of non-critical areas, and editing and adding more optimization targets through an optimization algorithm configuration module. Meanwhile, through an optimization algorithm configuration module, an optimization algorithm in a multi-objective optimization algorithm module based on a twin database can be designated, and the optimization algorithm is a genetic algorithm and a chebyshev decision (shown in fig. 1) in the invention. 4) The control task configuration is a specific parameter configuration of the control task in the digital twin world, such as task execution sequence, execution time limit and the like. The input of the module is strategy of strategy scheduling output, the output is specific execution task, and the specific execution task is issued to the entity object of the digital world for action execution.
4. And a multi-objective optimization algorithm module. The module mainly comprises an optimization target and an optimization algorithm, wherein the optimization target refers to three optimization functions in the invention: 1) Maximizing video quality in key areas; 2) Ensuring balance rate of overall network resource usage; 3) And the non-key area monitors the video quality. The optimization algorithm introduces a genetic algorithm and a chebyshev decision in the invention, and is responsible for synthesizing three optimization targets to obtain an optimal network resource allocation strategy and related scheduling parameters. For example, when network bandwidths of different monitoring areas need to be set so that three optimization targets are comprehensively optimal, the network bandwidths are taken as parameters, the optimal bandwidth parameters are output through iteration of an optimization algorithm, the bandwidth parameters are issued and bandwidth redistribution is executed, and then whether the current monitoring quality is optimized is evaluated. Note that in the present invention, both the optimization objectives and the optimization algorithm may be extended, for example, more optimization algorithms are introduced to enhance multi-objective optimization performance, more optimization objectives are introduced and configured, and so on. The expanding function can be configured and edited by engineers through a unified management platform.
5. And (5) a unified management platform. Through this platform, the engineer can regulate and control whole system, and specifically, unified management platform contains: database functions, information rights management functions, system optimization functions, visual interfaces, rule triggering functions, and security operation and maintenance functions. The database, the information authority management, the system optimization, the visualization and the security operation and maintenance are all conventional functions (refer to most of back-end systems), and the rule triggering function is related to the digital twin decision of the invention. Specifically, when the twin simulation environment executes the network resource scheduling policy output by the multi-objective optimization algorithm module, the rule triggering module is called first, and then whether the scheduling policy is executed is determined. For example, according to the current network environment, a rule is added in the rule triggering module, the ratio of network resources of a core area to network resources of a non-core area cannot be lower than 3 and cannot exceed 5, the temporary fault of the switch A cannot be involved in the process of executing scheduling, and the rules interfere simulation execution in the twin world and ensure that the execution accords with the current real world state.
In fig. 1, the overall execution flow of the system of the present invention is that, in the first step, physical world perception data is collected; secondly, constructing a simulation world by digital twin, wherein the simulation world can be adjusted according to configuration in the construction process; thirdly, requesting a multi-objective optimization algorithm module, and outputting an optimal scheduling strategy and related parameters according to the current state in a simulation environment so as to optimize the comprehensive optimization objective; step four, before executing the scheduling strategy, calling a rule configured by a unified management platform, determining that the current strategy is executable, if the current strategy conflicts with the rule configuration, adjusting the scheduling strategy configuration, and outputting the optimal strategy again based on a multi-objective optimization algorithm; and fifthly, after the optimal strategy is executed in the simulation environment, the monitoring performance is improved, and then the network resource scheduling strategy is fed back to the real physical world.
Specifically, referring to fig. 2 and 3, a multi-objective optimization algorithm in a multi-objective optimization algorithm module is combined with a chebyshev decision process to obtain an optimal network resource scheduling strategy; optimizing an objective function of three optimization objectives using a multi-objective optimization algorithm and chebyshev decision process, comprising:
S101: initializing a population: randomly generating a series of initial solutions as a population, wherein each solution represents a network resource scheduling scheme;
s102: objective function evaluation: for each solution in the population, three objective function values corresponding to the solution are calculated 、、；
S103: non-dominant ordering: non-dominant sorting is carried out on the population, and Pareto front grades of all solutions are determined;
S104: and (3) calculating the crowding degree: in each Pareto front, calculating the congestion degree of the solutions, and selecting solutions with low congestion degree so as to keep the diversity of the population;
S105: selection, crossover and mutation: applying the selection, crossover and mutation operations of the genetic algorithm to generate a new generation solution; in the selection process, individuals with non-dominant solutions and low crowding degrees are more prone to be selected;
s106: elite strategy: merging the non-dominant solution of the previous generation and the solution generated at present, and carrying out non-dominant sorting and congestion degree calculation again to ensure that elite solution is not lost;
s107: chebyshev decision process: for a Pareto optimal solution set found by an NSGA-III algorithm, performing weight distribution and selection on the Pareto optimal solution by using a Chebyshev method; setting a weight for each objective function, calculating the weighted objective function values to form Chebyshev functions, and selecting a solution which minimizes the Chebyshev functions as a final decision scheme;
S108: the iterative process: repeating S103 to S107 until a termination condition is satisfied; the termination condition includes reaching a maximum number of iterations or convergence of objective function values;
s109: and (3) outputting results: the solution that minimizes the chebyshev function is output as the optimal network resource scheduling policy.
S102 includes:
Wherein, For the maximization of the video quality of the key areas, K represents the set of all key areas,Is the weight of the critical area i,Is the video quality of region i under network resource allocation x;
Wherein, For the balance of overall network resource usage, N is the total number of areas monitored,Is the resource allocated to region i, T is the total network resource available;
Wherein, Is the minimum guarantee level of video quality for non-critical areas,Representing a set of all non-critical areas,Is the video quality of region j at network resource allocation x.
Resource allocation aimed at minimizing individual areasThe absolute difference from the average value, but taking into account that the monitoring of the critical area requires more resources, a weight coefficient can be introducedResource allocation for each region iWeighting is carried out; weight coefficientReflecting the importance of region i; setting a higher weight coefficient for the key area and setting a lower weight coefficient for the non-key area;
Involving weighting coefficients The modification is as follows:
When region i Lower than average resource allocationIf the region i is a key region, the multi-objective optimization algorithm tends to allocate more resources to the region i; otherwise, if the region i is a non-critical region, the multi-objective optimization algorithm allows the resources allocated to the region i to be lower than the average resources.
The expression of chebyshev functions is:
Wherein, Is an objective functionIs used for the weight of the (c),Is an objective functionIdeal values in the Pareto optimal solution set.
Referring to fig. 2, in an alternative embodiment, the NSGA-III multi-objective optimization algorithm is performed by the following steps:
S11: inputting network parameters, algorithm parameters and constraint variable coefficients; the network parameters comprise the network bandwidth allocation amount of each monitoring area, whether the switch starts link aggregation, whether port aggregation is started, and the upper limit and the lower limit of network resources of each monitoring area; the algorithm parameters comprise super parameters related to a genetic algorithm and a Chebyshev decision algorithm, an initial population size, a variation rate, a crossing rate, a maximum iteration number and the like of the genetic algorithm, a target weight, a target boundary, an optimization direction and the like of the Chebyshev decision algorithm; random generation of initial populations by genetic algorithm If (if)=1, Then S12 is performed; The iteration times;
S12: calculating a rapid non-dominant ranking fitness value, and carrying out non-inferior layering on the population; selecting a global optimal solution according to a crowding distance strategy; updating the position of the particles by using the P function to obtain a sub-population ；
S13: will be in the next generationAndIs combined into; Accumulating non-dominant ranking fitness value calculation and performing non-inferior layering; selecting a global optimal solution according to a crowding distance strategy; selecting the first N individuals to generate the next generation population；
S14: judging whether the maximum iteration times are reached, and if so, outputting an optimal Pareto solution; if it is not, make=+1; S12, switching to S12; until the maximum number of iterations is reached.
In an alternative embodiment, the method comprises the steps of:
1. Environment preparation and device deployment:
And installing a high-definition camera and sensor equipment in a key area of the hydropower plant.
The network device is configured to ensure a strong network connection to support the transfer of large amounts of data.
The server and storage device are installed to run the DSS and process large amounts of video data.
2. System configuration and initialization:
software tools required for installation and configuration on a server include database management systems, data processing software, and multi-objective optimization algorithms.
And constructing a digital twin model, and ensuring that the digital twin model reflects the accurate condition of the physical world.
Setting initial parameters and priority rules of a multi-objective optimization algorithm, and ensuring the consistency with the operation objective of the hydropower plant.
3. Data acquisition and transmission:
And starting the sensor and the camera, and collecting video data and environmental data of the monitoring area in real time.
The data is transmitted to the central processing server in real time through the network.
4. Data processing and event detection:
incoming data is processed and analyzed in real time, cleaned and converted to a suitable format.
An event detection algorithm is run to identify and classify significant events.
5. Multi-objective optimization and resource scheduling:
and dynamically distributing network resources based on the current network condition and the monitoring requirement by applying a multi-objective optimization algorithm.
And updating network resource allocation according to the algorithm result, and optimizing the transmission quality of video data.
6. Decision assistance and simulation:
And using the data analysis and simulation results provided by the DSS to assist the operation manager in making decisions.
Simulating the influence of network resource scheduling in a digital twin model, and predicting the results of different decision schemes.
7. Control execution and feedback:
The control parameters in the physical world are automatically or manually adjusted according to the DSS and the simulation results.
And monitoring the effect of control adjustment, and collecting feedback data for further optimizing the digital twin model and the multi-objective optimization algorithm.
8. Monitoring and maintaining:
The running state of the whole system is continuously monitored, including the use condition of network resources, the working state of a camera and the video quality.
System hardware and software are regularly maintained and upgraded to accommodate environmental changes and technological advances.
9. Security and backup:
And implementing a security policy to protect the monitoring data and the network communication.
Key data and system configuration are backed up periodically to cope with possible system failures or data loss.
The implementation of the present invention fully considers the challenges of practical industrial environments and provides a complete set of detailed steps to ensure stable and efficient operation of the system. Through intelligent network resource management, the high performance and the high reliability of the monitoring system of the hydropower plant are ensured, so that the safety monitoring capability and the operation efficiency of the hydropower plant are improved.
The invention aims at the algorithm designed for practical application in industry. The method is applied to a monitoring system, and combines a digital twin simulation technology and a multi-objective optimization algorithm to solve pain points of a current monitoring platform and balance quality and network resource allocation of video monitoring of different areas. The innovation points are as follows:
1. Multi-objective optimization: and dynamically balancing the quality of video monitoring in different areas and the network resource allocation by adopting a multi-target optimization algorithm.
2. And (3) real-time adjustment: and according to the real-time network condition and video quality feedback, automatically adjusting a resource allocation scheme, and ensuring the response speed and efficiency of the monitoring system.
3. Priority assessment: and implementing differentiated resource allocation strategies according to the importance degree and the security requirement of the region.
4. And (3) modular design: the modularized system design is easy to expand and maintain, and supports various network environments and video monitoring configurations.
Aiming at the situation that the video monitoring of the hydropower plant is limited in network resources and monitoring tasks of areas with different importance need to be processed, the invention ensures that the monitoring of the key areas is not affected by adjusting the network bandwidth allocation in real time, and simultaneously maintains the high availability and fault tolerance of the whole system. The system integrates a multi-objective optimization model and a Decision Support System (DSS), ensures that a key monitoring area obtains priority when network resources are tense through an algorithm, and provides real-time data visualization and intelligent early warning through the DSS so as to help operators to make timely and accurate decisions. In addition, the modular design of the system supports various network environments and video monitoring configurations, and wide applicability is ensured. The invention is characterized in that the multi-objective optimization algorithm is combined with the digital twin technology to simulate and verify the network resource scheduling scheme, and then the network resource scheduling scheme is implemented in the physical world, so that the economic and practical values of the video monitoring system of the hydropower plant are improved, and an innovative solution is provided for intelligent management and safety monitoring of the hydropower plant.
The invention adopts a multi-objective optimization algorithm to reasonably allocate network resources to each monitoring area, and ensure that important areas are monitored preferentially, thereby effectively solving the monitoring quality problem caused by insufficient network resources. By combining the algorithm model with the real-time monitoring data, the system can adjust network resources in real time, and high-definition video streams of a key monitoring area are ensured not to be interfered. Meanwhile, the invention simulates the optimization process in the virtual environment by introducing the digital twin technology so as to verify the network resource allocation scheme before physical world implementation. In addition, the integrated decision support system provides a highly visualized platform, and the intellectualization of operation management is further enhanced through real-time data display and intelligent early warning auxiliary decision making. The technical scheme of the invention provides an innovative network resource management method for real-time video monitoring in hydropower plants and even other industrial environments, optimizes monitoring efficiency and safety, and provides powerful technical support for development of industrial 4.0 and intelligent factories.
The invention provides a network resource real-time scheduling method based on a multi-objective optimization algorithm, which is used for dynamically distributing network resources of a hydropower plant monitoring system. The invention combines the multi-objective optimization and digital twin technology, realizes the real-time dynamic scheduling of network resources, ensures that the key area always maintains high-quality video monitoring, and simultaneously considers the effective utilization of the whole network resources and the quality assurance of non-key area monitoring.
The invention is based on the deep analysis of the existing hydropower plant video monitoring network resource allocation method and the investigation of actual demands, recognizes the defects of the prior art in the aspects of dynamic resource allocation and real-time optimization, and provides a corresponding solution. According to the invention, the stability and response speed of video monitoring can be improved, and the coping capability of emergency is enhanced, so that the safety management level is improved and the potential risk is reduced.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
Claims (6)
1. A network resource real-time scheduling method based on a multi-objective optimization algorithm is characterized by comprising the following steps:
The physical world collects real-time state data of the real monitoring system platform and flows the collected information into the CPS event model, the twin database and the database of the unified management platform respectively; the real-time state data comprise monitoring picture quality, network bandwidth occupation condition and network resource allocation condition of different monitoring areas;
The CPS event model takes the preprocessed data as input, transmits the preprocessed data to a virtual world, builds a three-dimensional simulation scene, and outputs an optimal scheduling strategy and configuration parameters required by a unified management platform according to the current state in the simulation environment so that an optimization target in a multi-target optimization algorithm is optimal; before executing the scheduling strategy, calling a rule configured by a unified management platform, determining that the current strategy is executable, and if the current strategy conflicts with the rule configuration, adjusting the scheduling strategy configuration, and outputting an optimal strategy again based on a multi-objective optimization algorithm; after the optimal strategy is executed in the simulation environment, feeding back the network resource scheduling strategy to the physical world;
the multi-objective optimization algorithm in the multi-objective optimization algorithm module is combined with the Chebyshev decision process to obtain an optimal network resource scheduling strategy; optimizing an objective function of three optimization objectives using a multi-objective optimization algorithm and chebyshev decision process, comprising:
step 101: initializing a population: randomly generating a series of initial solutions as a population, wherein each solution represents a network resource scheduling scheme;
step 102: objective function evaluation: for each solution in the population, three objective function values corresponding to the solution are calculated 、、；
Step 103: non-dominant ordering: non-dominant sorting is carried out on the population, and Pareto front grades of all solutions are determined;
step 104: and (3) calculating the crowding degree: in each Pareto front, calculating the congestion degree of the solutions, and selecting solutions with low congestion degree so as to keep the diversity of the population;
step 105: selection, crossover and mutation: applying the selection, crossover and mutation operations of the genetic algorithm to generate a new generation solution; in the selection process, individuals with non-dominant solutions and low crowding degrees are more prone to be selected;
Step 106: elite strategy: merging the non-dominant solution of the previous generation and the solution generated at present, and carrying out non-dominant sorting and congestion degree calculation again to ensure that elite solution is not lost;
Step 107: chebyshev decision process: for a Pareto optimal solution set found by an NSGA-III algorithm, performing weight distribution and selection on the Pareto optimal solution by using a Chebyshev method; setting a weight for each objective function, calculating the weighted objective function values to form Chebyshev functions, and selecting a solution which minimizes the Chebyshev functions as a final decision scheme;
step 108: the iterative process: repeating steps 103 to 107 until a termination condition is met; the termination condition includes reaching a maximum number of iterations or convergence of objective function values;
step 109: and (3) outputting results: outputting a solution which minimizes the chebyshev function as an optimal network resource scheduling strategy;
The step 102 includes:
Wherein, For the maximization of the video quality of the key areas, K represents the set of all key areas,Is the weight of the critical area i,Is the video quality of region i under network resource allocation x;
Wherein, For the balance of overall network resource usage, N is the total number of areas monitored,Is the resource allocated to region i, T is the total network resource available;
Wherein, Is the minimum guarantee level of video quality for non-critical areas,Representing a set of all non-critical areas,Is the video quality of region j under network resource allocation x;
Resource allocation aimed at minimizing individual areas The absolute difference from the average value, but taking into account that the monitoring of the critical area requires more resources, a weight coefficient can be introducedResource allocation for each region iWeighting is carried out; weight coefficientReflecting the importance of region i; setting a higher weight coefficient for the key area and setting a lower weight coefficient for the non-key area;
Involving weighting coefficients The modification is as follows:
When region i Lower than average resource allocationIf the region i is a key region, the multi-objective optimization algorithm tends to allocate more resources to the region i; otherwise, if the area i is a non-critical area, the multi-objective optimization algorithm allows the resources allocated to the area i to be lower than the average resources;
The expression of the chebyshev function is as follows:
Wherein, Is an objective functionIs used for the weight of the (c),Is an objective functionIdeal values in Pareto optimal solution set;
the NSGA-III multi-objective optimization algorithm comprises the following steps:
S11: inputting network parameters, algorithm parameters and constraint variable coefficients; the network parameters comprise the network bandwidth allocation amount of each monitoring area, whether the switch starts link aggregation, whether port aggregation is started, and the upper limit and the lower limit of network resources of each monitoring area; the algorithm parameters comprise super parameters related to a genetic algorithm and a Chebyshev decision algorithm, and the initial population size, the variation rate, the crossover rate, the maximum iteration number of the genetic algorithm, and the target weight, the target boundary and the optimization direction of the Chebyshev decision algorithm; random generation of initial populations by genetic algorithm If (if)=1, Then S12 is performed; The iteration times;
S12: calculating a rapid non-dominant ranking fitness value, and carrying out non-inferior layering on the population; selecting a global optimal solution according to a crowding distance strategy; updating the position of the particles by using the P function to obtain a sub-population ；
S13: will be in the next generationAndIs combined into; Accumulating non-dominant ranking fitness value calculation and performing non-inferior layering; selecting a global optimal solution according to a crowding distance strategy; selecting the first N individuals to generate the next generation population；
S14: judging whether the maximum iteration times are reached, and if so, outputting an optimal Pareto solution; if it is not, make=+1; S12, switching to S12; until the maximum number of iterations is reached.
2. The method for real-time scheduling of network resources based on a multi-objective optimization algorithm according to claim 1, wherein the CPS event model is responsible for preprocessing the raw collected data and fine-granularity classification, the data preprocessing includes normalizing and denoising the data, and the fine-granularity classification can divide the data into physical object events, control unit events, sensor events and execution device events; the physical object event refers to an action event of an entity object in the monitoring video; the control unit event refers to a real-time state event of a device in the physical world, and the sensor event refers to action event and state information executed by a sensor in the physical world; the executing device event refers to whether the device executing the control policy has executed the network bandwidth adjustment policy, and execution status information.
3. The method for real-time scheduling of network resources based on a multi-objective optimization algorithm according to claim 1, wherein a virtual simulation system in a virtual world constructs a three-dimensional simulation scene using data received from a CPS event model; the virtual world has four functions of connection, perception, decision and control, wherein the connection means mapping the real environment state of the physical world and feeding back the optimization strategy simulated in the virtual world to the physical world; the sensing means that real-time sensing data of the physical world is transmitted to the virtual world for three-dimensional reconstruction and simulation; the decision refers to a decision scheduling scheme obtained by the virtual world based on the CPS event model and the multi-objective optimization algorithm, and the decision scheduling scheme is fed back to the physical world after the virtual world is verified to be free of errors; control refers to the mapping of the virtual world itself to the physical world.
4. The network resource real-time scheduling method based on the multi-objective optimization algorithm according to claim 3, wherein the virtual world integrates four functional sub-modules: the system comprises a strategy scheduling module, a three-dimensional reconstruction module, an optimization algorithm configuration module and a control task configuration module;
the input of the strategy scheduling module is the current monitoring quality state information of the virtual world, the output is the scheduling strategy, the scheduling strategy is issued to the control task, and the control task configuration module issues the control instruction to the specific equipment for execution; the three-dimensional reconstruction configuration module is the functional configuration of the digital twin when the virtual world rendering and construction are carried out, the input of the three-dimensional reconstruction configuration module is the data which is output by the CPS model and used for constructing the twin model, the output is the configuration related information of the constructed twin body, and the three-dimensional reconstruction configuration module is completed based on the configurations when the three-dimensional simulation model is constructed; the optimization algorithm configuration is to select an optimization target and an optimization algorithm, and meanwhile, an optimization algorithm in a multi-target optimization algorithm module based on a twin database is designated through an optimization algorithm configuration module, and the control task configuration is specific parameter configuration of a control task in a digital twin world, wherein the specific parameter configuration comprises a task execution sequence and execution time limitation; the input of the control task configuration module is a strategy of strategy scheduling output, the output is a specific execution task, and the specific execution task is issued to the entity object of the virtual world for action execution.
5. The network resource real-time scheduling method based on the multi-objective optimization algorithm according to claim 1, wherein the multi-objective optimization algorithm module mainly comprises an optimization objective module and an optimization algorithm module, the optimization algorithm module introduces a genetic algorithm and chebyshev decision, and the optimization algorithm module is responsible for synthesizing three optimization objectives to obtain an optimal network resource allocation strategy and related scheduling parameters;
When the network bandwidths of different monitoring areas are required to be set, so that three optimization targets are comprehensively optimal, the network bandwidths at the moment are parameters, the optimal bandwidth parameters are output through iteration of an optimization algorithm, the bandwidth parameters are issued and bandwidth redistribution is executed, and then whether the current monitoring quality is optimized is evaluated.
6. The network resource real-time scheduling method based on the multi-objective optimization algorithm according to claim 1, wherein the whole system is regulated and controlled through a unified management platform, and the unified management platform comprises an implementation database module, an information authority management module, a system optimization module, a visualization module, a rule triggering module and a security operation and maintenance module; wherein the rule triggering function is related to digital twin decision; and when the twin simulation environment executes the network resource scheduling strategy output by the multi-objective optimization algorithm module, the rule triggering module is called, and then whether the scheduling strategy is executed is determined.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410675779.0A CN118250167B (en) | 2024-05-29 | 2024-05-29 | Network resource real-time scheduling method based on multi-objective optimization algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410675779.0A CN118250167B (en) | 2024-05-29 | 2024-05-29 | Network resource real-time scheduling method based on multi-objective optimization algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118250167A CN118250167A (en) | 2024-06-25 |
| CN118250167B true CN118250167B (en) | 2024-07-30 |
Family
ID=91561032
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410675779.0A Active CN118250167B (en) | 2024-05-29 | 2024-05-29 | Network resource real-time scheduling method based on multi-objective optimization algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118250167B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119537316A (en) * | 2024-12-04 | 2025-02-28 | 华远陆港智慧物流科技有限公司 | A data asset active management method and system based on DCMM |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118055078A (en) * | 2022-11-15 | 2024-05-17 | 河海大学 | Digital twin edge resource allocation algorithm based on multi-objective optimization |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111542078B (en) * | 2020-05-04 | 2022-09-16 | 南通大学 | A method for elastic resource allocation of core network control plane in NFV environment |
| WO2022139879A1 (en) * | 2020-12-24 | 2022-06-30 | Intel Corporation | Methods, systems, articles of manufacture and apparatus to optimize resources in edge networks |
| CN115858110A (en) * | 2022-11-14 | 2023-03-28 | 西南交通大学 | Multi-objective optimization strategy-based multi-level task scheduling method |
-
2024
- 2024-05-29 CN CN202410675779.0A patent/CN118250167B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118055078A (en) * | 2022-11-15 | 2024-05-17 | 河海大学 | Digital twin edge resource allocation algorithm based on multi-objective optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118250167A (en) | 2024-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI725744B (en) | Method for establishing system resource prediction and resource management model through multi-layer correlations | |
| CN111045828B (en) | Distributed edge calculation method based on distribution network area terminal and related device | |
| CN118590430A (en) | A network integrated dynamic resource routing management system and method | |
| CN118250167B (en) | Network resource real-time scheduling method based on multi-objective optimization algorithm | |
| CN118479315B (en) | Communication system and data transmission method based on elevator Internet of things | |
| Frantzén et al. | Digital-twin-based decision support of dynamic maintenance task prioritization using simulation-based optimization and genetic programming | |
| CN118628334B (en) | Image processing method based on computing network | |
| CN104038540A (en) | Method and system for automatically selecting application proxy server | |
| CN119316292B (en) | Policy selection method for node backup in multi-domain network | |
| CN119065818B (en) | A data communication system and method based on future network | |
| CN118729374A (en) | Dynamic hydraulic balance control method for heating | |
| CN119149231A (en) | Cloud computing resource scheduling method and system based on big data | |
| CN118606016A (en) | A task distribution method and device based on Manage and Worker | |
| CN120179414A (en) | Resource scheduling method, device and equipment | |
| CN117221246A (en) | Flow real-time self-adaptive scheduling method and system | |
| CN119892905A (en) | Micro-service scheduling method, device, equipment and storage medium under k8s cluster | |
| CN120258406A (en) | A power inspection and dispatching method and system based on artificial intelligence | |
| CN118101344B (en) | Transmission security identification system, method and medium for 5G message | |
| CN118887012A (en) | A transaction supervision system based on risk control rule engine | |
| CN110086662B (en) | Method for implementing demand definition network and network architecture | |
| CN120256143B (en) | Online resource management system and method based on AI large model | |
| CN117914005B (en) | Distribution network lean panoramic monitoring system and method | |
| LU508576B1 (en) | A real-time update management system for measuring control points based on digital campus | |
| CN120523151B (en) | Robot scheduling method and system based on collaborative control | |
| CN119484433B (en) | Bandwidth scheduling method and system for high-speed routing switching |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |