CN119476878B - An AGV-based intelligent material distribution management optimization method and system - Google Patents
An AGV-based intelligent material distribution management optimization method and systemInfo
- Publication number
- CN119476878B CN119476878B CN202510050495.7A CN202510050495A CN119476878B CN 119476878 B CN119476878 B CN 119476878B CN 202510050495 A CN202510050495 A CN 202510050495A CN 119476878 B CN119476878 B CN 119476878B
- Authority
- CN
- China
- Prior art keywords
- agv
- task
- delivery
- data
- distribution
- 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
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06316—Sequencing of tasks or work
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Computing Systems (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- Development Economics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an intelligent material distribution management optimization method and system based on an AGV, and relates to the technical field of AGV dispatching management, wherein the method comprises the steps of acquiring and analyzing factory operation data to obtain a material demand prediction result; the factory operation data comprise material demand data and AGV state data, a first AGV distribution task sequence is generated by utilizing a multi-objective optimization algorithm based on the factory operation data, a second AGV distribution task sequence is generated based on the material demand prediction result and the first AGV distribution task sequence, and the AGV is controlled to execute the material distribution task according to the second AGV distribution task sequence, and meanwhile task execution monitoring is conducted. According to the invention, the intelligent optimization of material distribution management is realized by combining an LSTM network to predict material demands, adopting a multi-objective optimization algorithm to generate task sequences and monitoring the execution state of the AGV in real time, the operation efficiency of the AGV distribution system is remarkably improved, the energy consumption is reduced, and the dynamic balance of equipment loads is realized.
Description
Technical Field
The invention relates to the technical field of AGV scheduling management, in particular to an intelligent material distribution management optimization method and system based on an AGV.
Background
With the deep development of intelligent manufacturing, the industry in the 4.0 era has put higher demands on material distribution management. AGV (Automated Guided Vehicle) is used as core equipment of the intelligent factory material distribution system, and the optimization of the dispatching management system directly influences the production efficiency and the operation cost. Traditional AGV scheduling management mainly adopts a heuristic algorithm based on rules or a simple real-time response strategy, and the method has a large limitation in processing complex and changeable material demands. In recent years, with the development of artificial intelligence technology, some researchers have begun to try to apply machine learning algorithms to the optimization of AGV scheduling, such as using reinforcement learning for path planning, using neural networks to predict material demands, and so on. However, these methods often fracture links such as material demand prediction, task planning, real-time scheduling, and the like, and lack a systematic overall optimization scheme.
The existing material demand prediction method mainly adopts traditional time sequence analysis or a simple statistical model, periodic characteristics and long-term dependence of material demands cannot be fully considered, prediction accuracy is insufficient, in a task planning stage, the existing algorithm usually only considers a single optimization target (such as distribution efficiency), but neglects multidimensional indexes such as energy consumption, load balance and the like, global optimization is difficult to achieve, in a third, in a real-time scheduling process, due to the fact that effective utilization and dynamic optimization capability of an AGV real-time state are lacking, the adaptability of the system is poor when the system faces emergency, distribution delay or resource waste is easy to be caused, and finally, an information island phenomenon generally exists in the existing system, the cooperativity among all functional modules is insufficient, and end-to-end intelligent optimization is difficult to achieve.
Therefore, there is an urgent need for an intelligent material distribution management optimization method that can effectively solve the above problems, so as to improve the overall efficiency of the AGV distribution system. The invention provides an intelligent material distribution management optimization method and system based on an AGV, which are innovated in multiple dimensions from material demand prediction, multi-objective task optimization to real-time monitoring and the like, and belong to the technical fields of intelligent manufacturing and AGV scheduling management.
Disclosure of Invention
The present invention has been made in view of the above-described problems.
Therefore, the invention provides an intelligent material distribution management optimization method and system based on an AGV, which can solve the problems in the background technology.
The intelligent material distribution management optimization method based on the AGVs comprises the following steps of obtaining and analyzing factory operation data to obtain material demand prediction results, wherein the factory operation data comprise material demand data and AGV state data, generating a first AGV distribution task sequence by using a multi-objective optimization algorithm based on the factory operation data, generating a second AGV distribution task sequence based on the material demand prediction results and the first AGV distribution task sequence, controlling the AGVs to execute material distribution tasks according to the second AGV distribution task sequence, and performing task execution monitoring.
The method for optimizing intelligent material distribution management based on the AGV comprises the following steps of classifying material demand data according to material types to obtain material classification data, extracting material demand cycle characteristics in the material classification data, training by adopting a long and short term memory network LSTM based on the material demand cycle characteristics to obtain a material demand prediction model, calculating AGV distribution capacity data based on the AGV state data, and generating a material demand prediction result based on the material demand prediction model, the material cycle characteristics and the AGV distribution capacity data.
The method for optimizing intelligent material distribution management based on the AGV comprises the following steps that material demand data comprise material types, quantity and emergency degree, AGV state data comprise current positions, electric quantity and load conditions, the extraction process of material demand periodic characteristics is specifically to analyze the material classification data through an autocorrelation function, find main periods, construct a periodic characteristic function and obtain material periodic characteristics based on the periodic characteristic function.
The method comprises the steps of constructing an objective function set based on factory operation data, determining conflict degrees among AGV delivery tasks based on the objective function set, determining task priorities according to the conflict degrees, executing a plurality of groups of co-evolution algorithms based on the objective function set and the task priorities, calculating comprehensive scores of each solution based on a global non-dominant solution set and the task priorities, selecting an optimal solution according to the comprehensive scores, and generating a first AGV delivery task sequence, wherein the first AGV delivery task sequence comprises AGV numbers, task numbers and execution time information.
As a preferable scheme of the intelligent material distribution management optimization method based on the AGV, the objective function set comprises a distribution efficiency objective function, an energy consumption objective function, a device load balancing objective function and an emergency objective function.
The conflict degree is comprehensively calculated through time difference, space distance and processing time difference, and the formula is expressed as follows:
;
in the formula, 、The method comprises the steps of respectively setting expected starting task time of an ith delivery task and a jth delivery task, wherein the ith delivery task and the jth delivery task form a task pair; The time difference for the largest expected start task between all delivery task pairs; The spatial distance between the ith delivery task and the jth delivery task is the spatial distance between the ith delivery task and the jth delivery task; maximum spatial distance between all pairs of delivery tasks; 、 the predicted processing time of the ith delivery task and the jth delivery task is respectively; The maximum processing time difference between all the distribution task pairs is obtained; 、、 The weight coefficients of the three dimensions of time difference, space distance and processing time are respectively adopted.
The task priority is determined by the ratio relation of the task urgency and the conflict degree:
;
in the formula, A priority value for the ith delivery task; The conflict degree sum of the ith delivery task and all other delivery tasks is given, and u i is the task urgency.
As a preferable scheme of the intelligent material distribution management optimization method based on the AGV, the invention comprises the following steps:
Constructing K sub-populations:
;
wherein, the Is the number of sub-populations;
For each of said sub-populations When the optimal solution is not updated in the continuous N generations, calculating the population diversity index:
;
wherein, the Scale for the ith sub-population; Is a sub-population Is a subject of (a); the number of objective functions; The evaluation value of the individual x for the jth objective function; for the jth objective function in the population Average value of (a); A diversity index value for the ith sub-population;
If it is Cross-population recombination is performed:
;
wherein, the For an individual from the current population,Individuals in a non-dominant solution set from other populations; is a new individual generated after recombination; Is a recombination weight coefficient; is a threshold value of the diversity index.
If it isThen adaptive mutation is performed:
;
wherein, the G is the current algebra,B is a shape parameter, which is the maximum algebra; x is the individual before mutation; is the upper bound of the decision variable; is the lower bound of the decision variable; In the form of a random number, ∈[0,1];Is an adaptive mutation step length.
And carrying out information exchange among populations every M generations, and updating the global non-dominant solution set through calculating the dominant relation of the solutions.
As a preferable scheme of the intelligent material distribution management optimization method based on the AGVs, the monitoring comprises AGV position tracking, task completion degree evaluation and abnormal situation report.
In order to further solve the technical problems, the intelligent material distribution management optimization system based on the AGVs comprises a data analysis module, a task generation module and a task optimization execution module, wherein the data analysis module is used for acquiring and analyzing factory operation data to obtain a material demand prediction result, the task generation module is used for generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data, the task optimization execution module is used for generating a second AGV distribution task sequence based on the material demand prediction result and the first AGV distribution task sequence, and the AGVs are controlled to execute material distribution tasks according to the second AGV distribution task sequence and simultaneously perform task execution monitoring.
A computer device comprising a memory and a processor, said memory storing a computer program, wherein said processor, when executing said computer program, implements the steps of an intelligent material distribution management optimization method based on an AGV as described above.
A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of an intelligent material distribution management optimization method based on an AGV as described above.
The invention has the beneficial effects that the intelligent optimization of material distribution management is realized by combining the LSTM network to predict the material demand, adopting the multi-objective optimization algorithm to generate the task sequence and monitoring the execution state of the AGV in real time, the operation efficiency of the AGV distribution system is obviously improved, the energy consumption is reduced and the dynamic balance of the equipment load is realized.
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, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is an overall flow chart of an intelligent material distribution management optimization method based on an AGV according to one embodiment of the invention.
FIG. 2 is a computer device diagram of the AGV-based intelligent material distribution management optimization method of the present invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Embodiment 1 referring to fig. 1, for one embodiment of the present invention, an intelligent material distribution management optimization method based on an AGV is provided.
FIG. 1 shows an overall flow chart of an intelligent material distribution management optimization method based on an AGV, comprising:
S1, acquiring and analyzing factory operation data to obtain a material demand prediction result;
s2, generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data;
And S3, generating a second AGV delivery task sequence based on the material demand prediction result and the first AGV delivery task sequence, controlling the AGV to execute the material delivery task according to the second AGV delivery task sequence, and simultaneously performing task execution monitoring.
In this embodiment, the details of S1 to S3 will be described one by one:
s1, acquiring and analyzing factory operation data to obtain a material demand prediction result.
Specifically, the factory operating data includes material demand data and AGV status data. Wherein the material demand data includes material type, quantity, and degree of urgency. The AGV status data includes current position, power and load conditions.
It should be noted that, in the invention, the acquisition of factory operation data is realized by adopting a layered data acquisition architecture, namely, material demand data is acquired in real time through intelligent data acquisition terminals and RFID technology deployed at each workstation, the current position, electric quantity and load conditions of an AGV are respectively acquired by utilizing a vehicle-mounted positioning system (such as laser SLAM or UWB positioning), a Battery Management System (BMS) and a weight sensor of the AGV, a reliable data transmission channel is constructed by adopting an industrial wireless network (such as an industrial Wi-Fi or 5G private network) and an OPC UA protocol, and a field data cache server is arranged to ensure the reliability of data transmission. In addition, it is recognized by those skilled in the art that real-time denoising, outlier detection and standardization processing are required to be performed on the collected raw data, and a data quality assessment mechanism is established to provide high-quality data support for subsequent data analysis and prediction models.
S101, classifying material demand data according to material types to obtain material classification data:
;
;
Wherein M is a material type set, n is the total number of material types; Classifying data for materials, specifically global historical demand data of the ith class of materials, T hist is the time span of the global historical demand data, and ~One of them can be expressed asMeaning the demand of the i-th material at a time point t;
It should be noted that when classifying the material demand data according to the material type, the classification rules are mainly based on the classification rules that the first class is classified according to physical properties of the material, such as taking liquid material, solid bulk material, single material, etc. as different types (m 1, m2, m 3.) because the classification rules are directly related to different carrier types that the AGV needs to be equipped with, the second class is classified according to storage conditions of the material, such as normal temperature material, constant temperature material, dangerous chemical, etc. because the specific conditions that the AGV needs to meet during distribution are determined, and the third class is classified according to usage of the material in production links, such as raw materials for production, auxiliary materials for packaging, equipment maintenance materials, etc., and the classification rules can reflect different timeliness requirements of the material demand. Through the multi-dimensional classification method, the data information of each material at each time point can be accurately recorded and analyzed, and more targeted data support is provided for subsequent demand prediction and AGV task planning.
S102, extracting material demand cycle characteristics by adopting a time sequence analysis method based on material classification data.
Specifically, data is classified from materialsExtracting the periodic characteristics of each type of material. First by an autocorrelation functionAnalyzing global historical demand data of material demands to find a primary periodAnd constructing a periodic characteristic function, and obtaining the periodic characteristic of the material based on the periodic characteristic function.
;
;
;
Wherein, the Is the autocorrelation function of the i-th material; k is a time lag value; Is the average value of the material demand of the i type; k is the length of the maximum detection period; The material periodic characteristics are output as periodic characteristic functions; 、 Is a parameter of the periodic characteristic function.
It should be noted that in the prior art, the material demand prediction often ignores the periodic characteristics of different materials, and the periodic fluctuation of the demand cannot be fully considered, so that the accuracy of the prediction is reduced when the material demand has a significant periodicity. The present invention thus introduces an autocorrelation functionThe method comprises the steps of extracting material demand periodic characteristics, constructing a periodic characteristic function to realize quantitative analysis of periodic fluctuation, providing accurate periodic input data for a subsequent material demand prediction model, and solving the problem that the periodic variation is difficult to accurately capture by time sequence analysis.
And S103, training by adopting a long-short-period memory network LSTM based on the material demand period characteristics to obtain a material demand prediction model.
In particular, using material cycle characteristicsAnd short-term historical demand dataAs input, an LSTM model is adopted for training, and a material demand prediction model is constructed. Capturing long-term dependency and periodic fluctuation characteristics in material demands by using LSTM model to generate original material demand predicted values at future time pointsThe formula is:
;
wherein, the Future generation of class i materials for LSTM model generationAn original demand forecast value of time; for LSTM model according to material period characteristics And short-term historical demand dataAnd (5) generating a demand prediction function.
It should be noted that,The method is characterized in that the method is used for representing short-term historical demand data, the historical demand data of the ith material in the last T time steps are used as the input of an LSTM model, and T is the time step length used for inputting the LSTM model. Short-term historical demand dataAnd global historical demand dataThe relationship between (i.e., material classification data) is that the short-term historical demand data is a time window extracted from the global historical demand data for input as a model at each prediction. This time window is typically a data segment of the last several time steps to reflect the current trends and changes.
Assuming global historical demand dataAnd a short-term historical demand data, of time length T hist, representing the complete demand data sequence from the initial point in time to the current timeThe time length of (2) is T, i.e. the time window length of the input LSTM model, and the short-term history data is extracted from the global history data according to the following method:
At time point t, the short-term historical demand data is expressed as:
;
A short-term window starting point is located. From global historical demand data Data from time T-T to T-1 are found.
Time window data is extracted. From global historical demand dataExtracting data of the last T time stepsConstitutes short-term historical demand data。
For example, if T hist = 100, representing that the global historical demand data contains 100 time steps, and T = 10, representing that the short-term time window length is 10, then at time point T = 100, the short-term historical demand data is:
。
In the prior art, conventional material demand prediction methods (such as linear regression or simple time series analysis) lack processing power for long-term dependence, and are difficult to adapt to material demand scenes with long-term dependence and complex fluctuations. The LSTM model can capture long-time dependence in a time sequence, and by combining LSTM and material periodic characteristics, the invention improves the recognition capability of long-term dependence and periodic variation, and solves the problem that the traditional model cannot consider long-term and short-term demand variation.
S104, calculating AGV delivery capacity data based on the AGV state data:
;
Wherein, the The AGV distribution capacity data at the time point t is used for representing the overall AGV distribution capacity; the efficiency coefficient of the jth AGV; the load capacity of the jth AGV; The current electric quantity of the jth AGV is obtained; full capacity for AGV; the distance from the jth AGV to the target position is set; Is the maximum delivery distance.
It should be noted that prior art AGV delivery capability assessment often uses static parameters that fail to account for changes in real-time conditions (e.g., power, load, and position), resulting in delivery capability that is inconsistent with the actual situation. Thus the invention calculatesThe invention has the advantages that the invention has dynamic adaptability in the evaluation of the delivery capacity, particularly can avoid the scheduling problem caused by insufficient delivery capacity under the conditions of frequent delivery tasks and large fluctuation of demands, and improves the responsiveness of the system.
S105, generating a material demand prediction result based on the material demand prediction model, the material cycle characteristics and the AGV delivery capacity data.
Specifically, AGV delivery capability data for the time point t and the original demand forecast generated in combination with the LSTM modelGenerating a final material demand prediction result. By balancing factorsAnd controlling the weight of the material demand prediction and the delivery capacity, and ensuring that the generated prediction result accords with the current actual delivery capacity of the AGV.
;
Wherein, the In the future for the i-th materialA demand forecast value of time; as a balancing factor for regulating LSTM prediction and And a weight therebetween.
It should be noted that prior art demand predictions are typically single-dimensional, i.e., generating material demand purely based on time series models (e.g., LSTM, ARIMA, etc.). However, the conventional model ignores the actual resource limitation (such as the AGV delivery capability) of the delivery system, so that the generated demand prediction result may exceed the actual delivery capability of the system, and the situations of overload of the system, delivery delay and even unbalance of resource scheduling are easy to occur. Especially in complex situations with large fluctuation of demand and limited distribution resources, the prediction results in the prior art cannot be directly converted into reasonable distribution plans. In order to solve the technical problems, the formula provided by the invention not only depends on the LSTM prediction result, but also calculates in real timeBy introducing the balance factor gamma between the demand predicted value and the actual delivery capacity as a constraint factor, the invention can dynamically adjust the weights of the demand predicted value and the actual delivery capacity under different conditions, and ensure that the predicted result meets the material demand change trend and accords with the bearing capacity of the actual delivery system. Thus, even under the condition of peak demand or limited resources, the system can generate more reasonable and executable prediction results based on the current resource conditions, and ensure that the prediction is suitable for the execution capacity.
And S2, generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data.
Specifically, the multi-objective optimization algorithm comprehensively considers a plurality of delivery indexes including delivery efficiency, energy consumption, equipment load and emergency degree.
S201, constructing an objective function set based on factory operation data.
Specifically, the objective function set includes a delivery efficiency objective function, an energy consumption objective function, a device load balancing objective function, and an emergency level objective function. The set of objective functions is expressed as:
;
wherein the delivery efficiency objective function is expressed as:
;
in the formula, Distance for the ith delivery task; The running speed when the ith delivery task is executed for the AGV; The material loading and unloading time of the ith delivery task, and n is the total number of delivery tasks.
The energy consumption objective function is expressed as:
;
in the formula, Is a load factor; Is unit energy consumption.
The device load balancing objective function is expressed as:
;
in the formula, The number of delivery tasks for the ith AGV.
The emergency level objective function is expressed as:
;
in the formula, Emergency coefficients for the delivery tasks; Is the response time.
S202, determining conflict degrees among AGV delivery tasks based on the target function set, and determining task priorities based on the conflict degrees.
Specifically, the conflict degree is obtained by comprehensively calculating a time difference, a space distance and a processing time difference, and the formula is as follows:
;
in the formula, 、The method comprises the steps of respectively setting expected starting task time of an ith delivery task and a jth delivery task, wherein the ith delivery task and the jth delivery task form a task pair; The time difference for the largest expected start task between all delivery task pairs; a spatial distance between the ith delivery task and the jth delivery task (i.e., a shortest path distance from a start point to an end point of the two delivery tasks); maximum spatial distance between all pairs of delivery tasks; 、 The predicted processing time (including running time and loading and unloading time) of the ith delivery task and the jth delivery task are respectively the task processing time; The maximum processing time difference between all the distribution task pairs is obtained; 、、 The weight coefficients of the three dimensions of time difference, space distance and processing time are respectively adopted.
The task priority is determined by the relation of the ratio of the emergency degree to the conflict degree of the delivery task, and the formula is as follows:
;
in the formula, A priority value for the ith delivery task; The sum of the conflict degrees of the ith delivery task and all other delivery tasks is calculated (namely, the sum of the conflict degrees of the ith delivery task and the remaining n-1 delivery tasks is calculated). The formula shows that the priority of a delivery task is proportional to its urgency and inversely proportional to its total conflict, i.e. the less conflicting delivery tasks will get a higher priority with other delivery tasks with the same urgency.
And S203, executing a plurality of group co-evolution algorithms based on the objective function set and the task priority.
The multi-population co-evolution algorithm comprises the steps of constructing K sub-populations, calculating a population diversity index for each sub-population, selecting to execute cross-population recombination or self-adaptive mutation operation according to the size of the population diversity index, exchanging information among the populations at intervals of preset algebra, and updating a global non-dominant solution set.
Specifically, first, K sub-populations are constructed:
;
wherein, the Is the number of sub-populations;
For each of said sub-populations When the optimal solution is not updated in the continuous N generations, calculating the population diversity index:
;
wherein, the Scale for the ith sub-population; Is a sub-population Is a subject of (a); the number of objective functions; The evaluation value of the individual x for the jth objective function; for the jth objective function in the population Average value of (a); A diversity index value for the ith sub-population;
If it is Cross-population recombination is performed:
;
wherein, the For an individual from the current population,Individuals from non-dominant solution sets of other populations; is a new individual generated after recombination; Is a recombination weight coefficient; is the diversity index value of the ith sub-population.
If it isThen adaptive mutation is performed:
;
wherein, the G is the current algebra,B is a shape parameter, which is the maximum algebra; x is the individual before mutation; is the upper bound of the decision variable; is the lower bound of the decision variable; In the form of a random number, ∈[0,1];Is an adaptive mutation step length.
And carrying out information exchange among populations every M generations, and updating the global non-dominant solution set through calculating the dominant relation of the solutions.
And S204, calculating the comprehensive score of each solution based on the global non-dominant solution set and the task priority, and selecting an optimal solution according to the comprehensive score to generate the first AGV delivery task sequence, wherein the first AGV delivery task sequence comprises an AGV number, a task number and execution time information.
And S3, generating a second AGV delivery task sequence based on the material demand prediction result and the first AGV delivery task sequence, controlling the AGV to execute the material delivery task according to the second AGV delivery task sequence, and simultaneously performing task execution monitoring.
Specifically, the monitoring includes AGV position tracking, task completion assessment and abnormal situation reporting.
In summary, the invention realizes the intelligent optimization of material distribution management by combining with the LSTM network to predict the material demand, adopting a multi-objective optimization algorithm to generate the task sequence and monitoring the execution state of the AGV in real time, thereby remarkably improving the operation efficiency of the AGV distribution system, reducing the energy consumption and realizing the dynamic balance of the equipment load.
The embodiment 2 provides an intelligent material distribution management optimization system based on an AGV, which comprises a data analysis module, a task generation module and a task optimization execution module, wherein the data analysis module is used for acquiring and analyzing factory operation data to obtain a material demand prediction result, the task generation module is used for generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data, the task optimization execution module is used for generating a second AGV distribution task sequence based on the material demand prediction result and the first AGV distribution task sequence, and controlling the AGV to execute a material distribution task according to the second AGV distribution task sequence, and meanwhile performing task execution monitoring.
Embodiment 3 referring to fig. 2, which is an embodiment of the present invention, it is different from the previous embodiment in that the functions, if implemented in the form of software functional units and sold or used as independent products, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium include an electrical connection (an electronic device) having one or more wires, a portable computer diskette (a magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of techniques known in the art, discrete logic circuits with logic gates for implementing logic functions on data signals, application specific integrated circuits with appropriate combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Embodiment 4, for one embodiment of the present invention, provides an intelligent material distribution management optimization method based on an AGV, and in order to verify the beneficial effects of the present invention, scientific demonstration is performed through economic benefit calculation and simulation experiments.
Experiments were performed at a large automobile parts manufacturing facility, which had 42 production workstations, 12 AGV robots. Production data of the factory 2023, 10 months and 2024, 3 months was selected as experimental base data. First, material demand data including a total of 84,632 distribution records of 314 different types of materials are collected within 6 months through an RFID data acquisition terminal deployed at each workstation. And cleaning the original data by adopting a data preprocessing module, and removing abnormal values to obtain 82,947 pieces of effective data. Subsequently, the data are divided into A, B, C classes based on the material type, wherein class a is the high frequency critical material (30% of total demand), class B is the medium frequency material (45% of total demand), and class C is the low frequency material (25% of total demand).
The improved autocorrelation function is utilized to analyze the demand cycle characteristics of various materials, the A-class materials mainly show the double cycle characteristics of 4 hours and 8 hours, the B-class materials show the single cycle characteristics of 12 hours, and the C-class materials show the long cycle characteristics of 24 hours. Based on these periodic characteristics, a three-layer LSTM network (input layer node 128, hidden layer node 256, output layer node 64) was constructed, trained using the first 5 months of data, and the last 1 month of data for testing. Meanwhile, position, electric quantity and load data are collected in real time through a sensor system carried on the AGV, the sampling frequency is 1Hz, and an AGV state database is constructed.
In the task optimization stage, four objective functions, namely a delivery efficiency objective function (taking the shortest completion time as an optimization target), an energy consumption objective function (taking the smallest total energy consumption as an optimization target), a device load balancing objective function (taking the smallest standard deviation of each AGV working load as an optimization target) and an emergency degree objective function (taking the highest emergency task priority as an optimization target), are designed. An improved multi-group co-evolution algorithm is adopted, the population scale is set to be 200, the evolution algebra is 1000 generations, the crossover probability is 0.8, and the mutation probability is 0.1. In the experiment, the traditional method, the single-target optimization method and the method of the invention are respectively adopted for comparison test, each group of test is repeated 30 times to obtain an average value, and the results are shown in table 1:
table 1 experimental data vs. table:
| Optimization method | Average delivery completion time (min) | Energy consumption index (kWh/h) | Load balancing degree of equipment (%) | Emergency task response time (min) | Accuracy of material delivery (%) | System fitness score |
| Traditional rule method | 45.8 | 12.4 | 68.5 | 15.6 | 92.3 | 0.72 |
| Single-objective optimization method | 38.2 | 10.8 | 72.4 | 12.3 | 94.8 | 0.78 |
| Group A of the invention | 28.4 | 8.6 | 89.7 | 6.8 | 98.2 | 0.91 |
| Group B of the invention | 29.1 | 8.9 | 88.5 | 7.2 | 97.8 | 0.90 |
| Group C of the invention | 27.9 | 8.4 | 90.2 | 6.5 | 98.5 | 0.92 |
| Group D of the invention | 28.6 | 8.7 | 89.1 | 6.9 | 98.1 | 0.91 |
;
The following analysis conclusion can be obtained by comparing experimental data, wherein the method is obviously superior to the traditional method and the single-target optimization method in all performance indexes. In terms of average delivery completion time, the method is shortened by about 39.1% compared with the traditional rule method and about 25.7% compared with the single-objective optimization method. This significant improvement benefits mainly from the synergistic effect of the LSTM network on accurate prediction of material demand and the multi-objective optimization algorithm. In terms of energy consumption indexes, the average energy consumption of the method is 8.65kWh/h, and is reduced by 30.2% compared with the traditional method, which is due to global consideration of AGV energy consumption in the task sequence optimization process.
In the aspect of equipment load balancing, the invention achieves the average level of 89.4 percent, improves the percentage point by 30.5 percent compared with the traditional method, and reflects the superiority of multi-objective optimization in task allocation. In the emergency task response time, the average response time of the invention is only 6.85 minutes, which is shortened by 56.1 percent compared with the traditional method, and the invention benefits from the dynamic adjustment mechanism of the task priority. The material distribution accuracy reaches a high level of 98.2%, and is improved by 5.9% compared with the traditional method, so that the high-precision characteristic of the prediction model is reflected.
The system adaptability score is a comprehensive evaluation index and comprises a plurality of dimensions such as the processing capacity of the system for emergency, the task dynamic adjustment capacity and the like. The average score of the invention is 0.91, which is improved by 26.4% compared with the traditional method, which shows that the method has stronger environmental adaptability and robustness. Through the analysis of variance of A, B, C, D groups of experimental data, when the confidence coefficient is 95%, the standard deviation of each index is less than 3%, and the invention has good stability and repeatability. These experimental data fully demonstrate the innovative and practical value of the present invention in the field of intelligent material distribution management optimization.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (7)
1. An intelligent material distribution management optimization method based on an AGV is characterized by comprising the following steps:
Acquiring and analyzing factory operation data to obtain a material demand prediction result, wherein the factory operation data comprises material demand data and AGV state data;
generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data;
generating a second AGV delivery task sequence based on the material demand prediction result and the first AGV delivery task sequence, controlling an AGV to execute a material delivery task according to the second AGV delivery task sequence, and simultaneously performing task execution monitoring;
the process for obtaining the material demand prediction result comprises the following steps:
Classifying the material demand data according to the material type to obtain material classification data;
Extracting material demand cycle characteristics in the material classification data;
Training by adopting a long-short-period memory network LSTM based on the material demand period characteristics to obtain a material demand prediction model;
calculating AGV delivery capacity data based on the AGV state data;
generating a material demand prediction result based on the material demand prediction model, the material cycle characteristics and the AGV distribution capacity data;
The method for generating the first AGV distribution task sequence by utilizing the multi-objective optimization algorithm based on the factory operation data comprises the following steps:
constructing an objective function set based on the plant operation data;
Determining conflict degrees among AGV delivery tasks based on the objective function set, and determining task priorities according to the conflict degrees;
Executing a plurality of group co-evolution algorithms based on the set of objective functions and the task priority;
Selecting an optimal solution according to the comprehensive score, and generating a first AGV delivery task sequence, wherein the first AGV delivery task sequence comprises an AGV number, a task number and execution time information;
The objective function set comprises a distribution efficiency objective function, an energy consumption objective function, a device load balancing objective function and an emergency degree objective function;
the conflict degree is comprehensively calculated through time difference, space distance and processing time difference, and the formula is expressed as follows:
;
in the formula, 、The method comprises the steps of respectively setting expected starting task time of an ith delivery task and a jth delivery task, wherein the ith delivery task and the jth delivery task form a task pair; The time difference for the largest expected start task between all delivery task pairs; The spatial distance between the ith delivery task and the jth delivery task is the spatial distance between the ith delivery task and the jth delivery task; maximum spatial distance between all pairs of delivery tasks; 、 the predicted processing time of the ith delivery task and the jth delivery task is respectively; The maximum processing time difference between all the distribution task pairs is obtained; 、、 The weight coefficients of three dimensions, namely time difference, space distance and processing time;
the task priority is determined by the ratio relation of the task urgency and the conflict degree:
;
in the formula, A priority value for the ith delivery task; The conflict degree sum of the ith delivery task and all other delivery tasks is given, and u i is the task urgency.
2. The intelligent AGV-based material distribution management optimization method according to claim 1 wherein said material demand data comprises material type, quantity and urgency;
The AGV state data comprises a current position, electric quantity and load conditions;
The extraction process of the material demand cycle characteristics specifically comprises the steps of analyzing the material classification data through an autocorrelation function, finding out a main cycle, constructing a cycle characteristic function, and obtaining the material cycle characteristics based on the cycle characteristic function.
3. The intelligent material distribution management optimization method based on AGV according to claim 2 wherein said plurality of group co-evolution algorithms comprises:
Constructing K sub-populations:
;
wherein, the Is the number of sub-populations;
For each of the sub-populations, when its optimal solution is not updated for N consecutive generations, calculating:
;
wherein, the Scale for the ith sub-population; Is a sub-population Is a subject of (a); the number of objective functions; An evaluation value of the kth objective function on the individual x; in population for the kth objective function Average value of (a); A diversity index value for the ith sub-population;
If it is Cross-population recombination is performed:
;
wherein, the For an individual from the current sub-population,Individuals in a non-dominant solution set from other sub-populations; is a new individual generated after recombination; Is a recombination weight coefficient; a threshold value that is a diversity indicator;
If it is Then adaptive mutation is performed:
;
wherein, the G is the current algebra,B is a shape parameter, which is the maximum algebra; x is the individual before mutation; is the upper bound of the decision variable; is the lower bound of the decision variable; In the form of a random number, ∈[0,1];Is the self-adaptive variation step length;
and carrying out information exchange among populations every M generations, and updating the global non-dominant solution set through calculating the dominant relation of the solutions.
4. The intelligent AGV-based material delivery management optimization method of claim 3 wherein said monitoring comprises AGV position tracking, task completion assessment and abnormal situation reporting.
5. A system employing the intelligent AGV-based material delivery management optimization method according to any one of claims 1 to 4, comprising:
the data analysis module is used for acquiring and analyzing factory operation data to obtain a material demand prediction result;
the task generation module is used for generating a first AGV distribution task sequence by utilizing a multi-objective optimization algorithm based on the factory operation data;
And the task optimization execution module is used for generating a second AGV delivery task sequence based on the material demand prediction result and the first AGV delivery task sequence, controlling the AGV to execute the material delivery task according to the second AGV delivery task sequence, and simultaneously performing task execution monitoring.
6. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor when executing the computer program performs the steps of the intelligent AGV-based material delivery management optimization method of any one of claims 1-4.
7. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the AGV-based intelligent material delivery management optimization method of any of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510050495.7A CN119476878B (en) | 2025-01-13 | 2025-01-13 | An AGV-based intelligent material distribution management optimization method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510050495.7A CN119476878B (en) | 2025-01-13 | 2025-01-13 | An AGV-based intelligent material distribution management optimization method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119476878A CN119476878A (en) | 2025-02-18 |
| CN119476878B true CN119476878B (en) | 2025-08-29 |
Family
ID=94578061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510050495.7A Active CN119476878B (en) | 2025-01-13 | 2025-01-13 | An AGV-based intelligent material distribution management optimization method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119476878B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120181536B (en) * | 2025-05-21 | 2025-08-08 | 福建省唯正智能科技有限公司 | AGV material circulation scheduling system of 12-axis PCB drilling machine |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117114313A (en) * | 2023-08-23 | 2023-11-24 | 江苏科技大学 | An AGV group scheduling method based on demand task prediction model |
| CN119200616A (en) * | 2024-10-09 | 2024-12-27 | 重庆工程职业技术学院 | An AGV control system for intelligent logistics |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116228284A (en) * | 2022-09-06 | 2023-06-06 | 广东京邦达供应链科技有限公司 | Goods demand prediction method, training device, computer system and medium |
| CN116401951A (en) * | 2023-04-10 | 2023-07-07 | 中国电子科技集团公司第三十八研究所 | Integrated multi-AGV path planning and multi-row equipment layout optimization method |
-
2025
- 2025-01-13 CN CN202510050495.7A patent/CN119476878B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117114313A (en) * | 2023-08-23 | 2023-11-24 | 江苏科技大学 | An AGV group scheduling method based on demand task prediction model |
| CN119200616A (en) * | 2024-10-09 | 2024-12-27 | 重庆工程职业技术学院 | An AGV control system for intelligent logistics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119476878A (en) | 2025-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115760484B (en) | Method, device and system for improving hidden danger identification capability of power distribution area and storage medium | |
| CN119476878B (en) | An AGV-based intelligent material distribution management optimization method and system | |
| CN118628161B (en) | Supply chain demand prediction method and system | |
| CN114154716B (en) | Enterprise energy consumption prediction method and device based on graph neural network | |
| CN117952009A (en) | Intelligent production line testable digital twin modeling method | |
| CN108491982A (en) | A kind of short-term load forecasting method and system based on echo state network | |
| CN113743673A (en) | Power load prediction method during typhoon | |
| CN112734135A (en) | Power load prediction method, intelligent terminal and computer readable storage medium | |
| CN112990587A (en) | Method, system, equipment and medium for accurately predicting power consumption of transformer area | |
| CN120146502B (en) | Business management method based on one-stop inspection platform | |
| CN120087722A (en) | An Optimization Method for Economic Resource Management Based on Intelligent Decision-making | |
| CN116470491A (en) | Photovoltaic power probability prediction method and system based on copula function | |
| CN115185247B (en) | A method for predicting machine tool operation status and workshop scheduling based on product processing quality | |
| CN119831293A (en) | Intelligent distribution system and method for dairy product production line | |
| CN112734141B (en) | Diversified load interval prediction method and device | |
| CN112330017B (en) | Power load forecasting method, device, electronic device and storage medium | |
| CN113902181A (en) | Short-term prediction method and equipment for heavy overload of public transformer | |
| CN116826745B (en) | Layered and partitioned short-term load prediction method and system in power system background | |
| CN113837486B (en) | RNN-RBM-based distribution network feeder long-term load prediction method | |
| CN120525140B (en) | Lean operation method based on production scheduling optimization and equipment OEE monitoring | |
| CN120386278B (en) | Intelligent decision method and system based on integrated industrial PLC controller of computational network | |
| CN119765253B (en) | Power grid load prediction method and system based on artificial intelligence and subspace aggregation | |
| CN120069615B (en) | Power transaction trend prediction and decision support system and method | |
| CN120073660A (en) | Virtual power plant load prediction control method and system based on big data | |
| CN120033661A (en) | Multi-load access optimization method and system in high reliability distribution network planning |
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 |