Research And Application Of Multi-objective Flexible Shop Scheduling Problem Based On Improved NSGA2 Algorithm

China has become the largest manufacturing country in the world since 2010,the rapid development of manufacturing industry is a sharp weapon to promote the development of real economy.The development of manufacturing industry is inseparable from intelligent manufacturing.Job shop scheduling is one of the important components of intelligent manufacturing,reasonable job shop scheduling scheme can improve production efficiency and delivery rate,as well as reducing production cost and energy consumption.The performance of the algorithm used in job shop scheduling determines the quality of job shop scheduling scheme,so the rapid development of manufacturing industry is closely related to intelligent scheduling methods.Elite non dominated sorting genetic algorithm（NSGA2）is of great value to the study of the multi-objective flexible job shop scheduling problem,but the randomization of evolutionary parameters affects the performance of the algorithm.Therefore,this paper focuses on the problem of the multi-objective flexible job shop scheduling problem,designs an NSGA2 algorithm based on reinforcement learning,uses the self-learning characteristics of reinforcement learning to improve the performance of the algorithm,and solves the multi-objective flexible job shop scheduling problem efficiently.The main innovative work of this paper is as follows:（1）For multi-objective optimization problems,an elite non dominated sorting genetic algorithm based on reinforcement learning is proposed by using the self-learning characteristic of reinforcement learning.The self-learning strategy of algorithm evolution parameters is designed so that the evolution parameters can be selected according to the current population state;secondly,the BAS algorithm was used to improve the speed of individual optimization and global search ability,so as to improve the global search ability of the algorithm and avoid algorithm falling into local optimization;finally,the classical multi-objective test function is used to test the algorithm,which compared with the traditional multi-objective algorithm,the results show that the proposed algorithm is effective in dealing with the multi-objective function optimization problem.（2）For multi-objective flexible job shop scheduling problem,the proposed Improved NSGA2 algorithm is applied to multi-objective flexible job shop scheduling problem.Firstly,according to the characteristics of flexible job shop scheduling problem and scheduling rules,a population initialization method is designed to improve the quality of the initialization population,and the crossover and mutation methods suitable for job shop scheduling problem are designed;secondly,by using the standard example of the problem to test the static scheduling and dynamic scheduling,and compared with other algorithms.The results show that the algorithm has the advantages and stability in solving the problem;finally,it was combined with the specific production case of the enterprise by the actual production situation of the products produced in the workshop.The constraints were added between the tasks and prove the practicability of the algorithm.