A method for implementing AGV scheduling in manufacturing workshop based on NSGA-II
By simplifying the AGV scheduling problem to a flexible work workshop problem and adopting the NSGA-II algorithm, using the codec method and fast non-dominant sorting, the complexity and inefficiency of AGV scheduling in the manufacturing workshop are solved, and efficient multi-objective optimization is achieved and scheduling scheme selection.
Patent Information
- Application Number
- CN202310121830.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-02-15
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2043-02-15
AI Technical Summary
The prior art is difficult to effectively solve the complexity and inefficiency of AGV scheduling problems in manufacturing workshops, especially in the selection of scheduling solutions under multi-objective optimization and multiple optimization indicators.
The AGV scheduling problem is simplified to a flexible work workshop problem, and the NSGA-II algorithm is used for optimization. Through the workpiece and machine-based encoding and decoding method, combined with fast non-dominant sorting and congestion comparison, an optimal scheduling solution is generated.
It improves the solution efficiency of AGV scheduling problems, takes into account multiple optimization goals, and selects the most appropriate scheduling solution in real time, reducing the complexity of the problem and the computational complexity.
Smart Images

Figure CN115903716B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent control technology, and in particular to a method for implementing AGV scheduling based on NSGA-II. Background Art
[0002] With the widespread application of emerging information technology, intelligent AGV transportation has developed rapidly; AGV can be used for horizontal transportation equipment, manufacturing workshops, catering, etc. in automated container terminals. Its efficient operation can improve the loading and unloading efficiency of all walks of life; AGV (Automated Guided Vehicle) means "automatic guided vehicle", which is equipped with electromagnetic or optical automatic guidance devices, which can travel along the specified guidance path and has safety protection and various transfer functions.
[0003] AGV, also known as unmanned guided vehicle, automatic guided vehicle, and laser guided vehicle, is characterized by its unmanned driving. The AGV is equipped with an automatic guidance system, which can ensure that the system can automatically travel along the predetermined route without the need for manual navigation, and automatically transport goods or materials from the starting point to the destination.
[0004] NSGA-II is one of the most popular multi-objective genetic algorithms. It reduces the complexity of the non-inferior sorting genetic algorithm and has the advantages of fast running speed and good convergence of the solution set, becoming the benchmark for the performance of other multi-objective optimization algorithms. It adopts a fast non-dominated sorting algorithm, and its computational complexity is greatly reduced compared with NSGA. It adopts crowding and crowding comparison operators instead of the shared radius shareQ that needs to be specified, and uses them as the winning criteria in the same-level comparison after quick sorting, so that individuals in the quasi-Pareto domain can be expanded to the entire Pareto domain and evenly distributed, maintaining the diversity of the population. It introduces an elite strategy to expand the sampling space, prevent the loss of the best individuals, and improve the algorithm's computing speed and robustness. Summary of the invention
[0005] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II.
[0006] The present invention is implemented by the following technical scheme: a method for realizing AGV scheduling in a manufacturing workshop based on NSGA-II, characterized in that: the workshop AGV scheduling problem is simplified to a flexible job shop problem, and is solved by a machine method and a workpiece-based encoding and decoding method, and is optimized by an NSGA-II method to finally obtain the optimal scheduling solution. The specific encoding and decoding steps are as follows:
[0007] Step 1: Problem coding: Encode the workpiece to be processed according to its process content using the workpiece-based coding method; compare all AGVs to optional machines in the "flexible job shop problem" and encode the AGV using the machine-based coding method;
[0008] Step 2: Generate a sequence: According to the encoding method, randomly generate a sequence for all the processes of all workpieces (all the processes to be decoded, after decoding, it becomes a feasible scheduling plan);
[0009] Step 3, decoding (generating the sequence of AGV scheduling): According to the coding information of the workpiece sequence, the optional AGVs are determined in the AGV set in turn, and finally an AGV is randomly selected and arranged on the coded machine sequence of the corresponding workpiece, and finally a complete machine sequence is generated, that is, the optional scheduling plan, which can be called a population individual;
[0010] Step 4: Calculate the value of the optimization target according to the machine sequence as the basis for evaluating the fitness of the individuals in the population.
[0011] The optimization steps based on NSGA-II are as follows:
[0012] Step 1: NSGA-II parameter setting: algorithm iteration number Gen max , population size N, selection probability P S , crossover probability P C , mutation probability P m , the maximum number of Pareto solution sets MaxSize;
[0013] Step 2, matching setting: Initialize N individuals to form a population P, the matching setting includes encoding of the workpiece and AVG and optimizing the target value;
[0014] Step 3: Numerical calculation: Let Gen = 0, and perform fast non-dominated sorting and crowding distance calculation based on f1 and f2 of each individual in population P;
[0015] Step 4: Update the Pareto solution set: According to the level and crowding distance of individuals in the population, store MaxSize individuals into the Pareto solution set. If there are already individuals in the Pareto solution set, reorder the current MaxSize individuals and the existing individuals, remove individuals with poor adaptability, and ensure that there are MaxSize individuals in the Pareto solution set.
[0016] Step 5: Form an ecosystem: Use the tournament selection method to give priority to individuals with the lowest non-dominated level, and then select larger individuals according to the crowding distance until the number of selections of population individuals is met. Select the individuals with the lowest non-dominated level to directly enter the next generation of population Q, perform crossover and mutation on other individuals, and finally generate population Q.
[0017] Step 6: The individuals of the population Q and the individuals of the population P are merged into a population QP, and are sorted from strong to weak in terms of fitness according to the optimization target value using the fast non-dominated sorting method, and the first N individuals with strong fitness are selected to form a new population P.
[0018] Step 7: Let Gen = Gen + 1; if Gen reaches the set maximum number of iterations, directly output the current optimal Pareto solution set, otherwise enter the next loop.
[0019] Preferably, the encoding of the workpiece and AVG is to encode the workpiece and AGV according to the process-based and machine-based encoding methods respectively; the processes of all workpieces are randomly arranged in the same sequence to generate a process sequence; for each process in the process sequence, an AGV is randomly selected from the AGV set to generate an AGV sequence in sequence.
[0020] Preferably, the optimization target value is to arrange the processes at each gene position in the process sequence to each AGV for transportation in turn, and calculate the optimization target value to form a population of individuals, repeat N times, and finally form a population P.
[0021] Preferably, the individuals of the population Q and the individuals of the population P are merged into a population QP, and fitness is evaluated by a fast non-dominated sorting method, and the individuals are sorted from strong to weak in terms of fitness, and the first N individuals are selected to form a new population P.
[0022] The beneficial effects of the present invention are: the differences and advantages of this method over the prior art are:
[0023] Compared with the existing traditional AGV technology: the use of flexible job shop problem model simplifies the complex AGV scheduling problem, and the mature encoding and decoding method can be used to encode and decode the AGV problem, thereby improving the solution efficiency and reducing the complexity of the problem itself.
[0024] Flexible application of NSGA-II technology: Using the fast non-dominated sorting method in NSGA-II, a non-inferior solution that takes into account multiple optimization objectives can be found, taking into account various optimization indicators, and selecting the most appropriate scheduling plan in real time according to the AGV transportation needs. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 The figure is a schematic diagram of the overall process of a method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II.
[0026] Figure 2 A schematic diagram of a method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II based on process coding and machine coding.
[0027] Figure 3 A schematic diagram of a decoded scheduling scheme for a method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II.
[0028] Figure 4 The figure is a schematic diagram of the actual usage status of a method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II. DETAILED DESCRIPTION
[0029] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0030] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments of the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.
[0031] A method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II can be used as a simplification of the workshop AGV scheduling problem into a flexible job shop problem, and is solved by a machine method and a workpiece-based encoding and decoding method. The NSGA-II-based method is used for optimization, and the optimal scheduling solution is finally obtained. The specific encoding and decoding steps are as follows:
[0032] Step 1: Problem coding: Encode the workpiece to be processed according to its process content using the workpiece-based coding method; compare all AGVs to optional machines in the "flexible job shop problem" and encode the AGV using the machine-based coding method;
[0033] Step 2: Generate a sequence: According to the encoding method, randomly generate a sequence for all the processes of all the workpieces. The purpose is to decode all the processes to be decoded into a feasible scheduling plan;
[0034] Step 3: Decoding: Generate the AGV scheduling sequence. According to the coding information of the workpiece sequence, determine the optional AGVs in the AGV set in turn. Finally, randomly select an AGV and arrange it on the coded machine sequence of the corresponding workpiece. Finally, generate a complete machine sequence, that is, the optional scheduling plan, which can be called a population individual.
[0035] Step 4: Calculate the value of the optimization target according to the machine sequence as the basis for evaluating the fitness of the individuals in the population.
[0036] The optimization steps based on NSGA-II are as follows:
[0037] Step 1: NSGA-II parameter setting: algorithm iteration number Gen max , population size N, selection probability P S , crossover probability P C , mutation probability P m , the maximum number of Pareto solutions MaxSize; set relevant parameters and specify the encoding of the target individual;
[0038] Step 2: Matching setting: Initialize N individuals to form a population P. The matching setting includes encoding the workpiece and AVG and optimizing the target value.
[0039] Step 3: Numerical calculation: Let Gen = 0, and perform fast non-dominated sorting and crowding distance calculation based on f1 and f2 of each individual in population P;
[0040] Step 4: Update the Pareto solution set: According to the level and crowding distance of individuals in the population, store MaxSize individuals into the Pareto solution set. If there are already individuals in the Pareto solution set, reorder the current MaxSize individuals and the existing individuals, remove individuals with poor adaptability, and ensure that there are MaxSize individuals in the Pareto solution set.
[0041] Step 5: Form an ecosystem: Use the tournament selection method to give priority to individuals with the lowest non-dominated level, and then select larger individuals according to the crowding distance until the number of selections of population individuals is met. Select the individuals with the lowest non-dominated level to directly enter the next generation of population Q, perform crossover and mutation on other individuals, and finally generate population Q.
[0042] Step 6: The individuals of population Q and population P are merged into population QP, and sorted from strong to weak in terms of fitness according to the optimization target value using the fast non-dominated sorting method, and the first N individuals with strong fitness are selected to form a new population P.
[0043] Step 7: Let Gen = Gen + 1; if Gen reaches the set maximum number of iterations, directly output the current optimal Pareto solution set, otherwise enter the next loop.
[0044] For the coding of workpieces and AVGs, the workpieces and AGVs are encoded according to the process-based and machine-based coding methods respectively; the processes of all workpieces are randomly arranged in the same sequence to generate a process sequence; for each process in the process sequence, an AGV is randomly selected from the AGV set to generate an AGV sequence in sequence.
[0045] Optimize the target value, arrange the processes at each gene position in the process sequence to each AGV for transportation, and calculate the optimized target value to form a population of individuals, repeat N times, and finally form a population P.
[0046] The individuals of population Q and population P are merged into population QP, and the fitness is evaluated by the fast non-dominated sorting method. The individuals are sorted from strong to weak in terms of fitness, and the first N individuals are selected to form a new population P.
[0047] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for implementing AGV scheduling in a manufacturing workshop based on NSGA-II. Features: It includes problem encoding, sequence generation, decoding, fitness evaluation optimization and NSGA-II-based optimization method; the specific steps are as follows: Step 1: Problem coding: Encode the workpiece to be processed according to its process content and the workpiece-based coding method; regard all AGVs as optional machines, encode the AGVs using the machine-based coding method to obtain the workpiece code and AGV code; Step 2: Generate sequence: According to the encoding method, randomly generate a sequence for all the workpiece processes to obtain the encoding information of the workpiece sequence; Step 3: Decoding: According to the coding information of the workpiece sequence, select the optional AGVs in the AGV set in turn, and finally randomly select an AGV and arrange it on the coded machine sequence of the corresponding workpiece to generate a complete machine sequence; The specific steps of the NSGA-II-based optimization method are as follows: Step 1: NSGA-II parameter setting: algorithm iteration number Gen max , population size N, selection probability P S , crossover probability P C , mutation probability P m , the maximum number of Pareto solution sets MaxSize; Step 2, matching setting: Initialize N individuals to form a population P, the matching setting includes encoding of the workpiece and AVG and optimizing the target value; Step 3: Numerical calculation: Let Gen = 0, and perform fast non-dominated sorting and crowding distance calculation based on f1 and f2 of each individual in population P; Step 4: Set Gen = Gen + 1; when Gen reaches the maximum number of iterations, directly output the current optimal Pareto solution set; The encoding of the workpiece and AVG is to encode the workpiece and AGV according to the process-based and machine-based encoding methods respectively; the processes of all workpieces are randomly arranged in the same sequence to generate a process sequence; for each process in the process sequence, an AGV is randomly selected from the AGV set to generate an AGV sequence in sequence; The optimization target value is to arrange the processes at each gene position in the process sequence to each AGV for transportation, and calculate the optimization target value to form a population individual, repeat N times, and finally form a population P; The machine sequence calculates the value of the optimization target as the basis for evaluating the fitness of the individuals in the population; The Pareto solution set is processed, the set of the Pareto solution set is updated, and according to the level and crowding distance of individuals in the population, MaxSize individuals are stored in the Pareto solution set. If there are individuals in the Pareto solution set, the current MaxSize individuals and the existing individuals are reordered, and individuals with poor adaptability are removed to ensure that there are MaxSize individuals in the Pareto solution set; the level and crowding distance of individuals in the population are selected by the tournament selection method, and individuals with the lowest non-dominated level are selected first, and then larger individuals are selected according to the crowding distance until the number of selections of individuals in the population is met. The individuals with the lowest non-dominated level are selected to directly enter the next generation population Q, and other individuals are crossovered and mutated to generate population Q; the individuals of the population Q are merged with the individuals of the population P into population QP, and are sorted from strong to weak according to the fitness ability according to the fast non-dominated sorting method, according to the optimization target value, and the top N individuals with strong fitness are selected to form a new population P.
