Research On Flexible Job-shop Scheduling Problem Based On Hybrid Intelligent Optimization Algorithm

Flexible Job-shop Scheduling Problem(FJSP)is a kind of widely used combinatorial optimization problem.A reasonable scheduling scheme can effectively improve the production efficiency,reduce costs and improve the robustness of the production system.Its application fields include assembly line processing,transportation,airport and port scheduling,medical and health care,etc.In addition,this paper considers the current hot issue of Distributed Flexible Job-shop Scheduling Problem(DFJSP).This problem aims at the integrated scheduling of the system with multiple flexible manufacturing units,which can plan the production system in a more complex dimension.However,based on the no free lunch theorem,any single algorithm can not guarantee that it can deal with all optimization problems perfectly in a limited time.Therefore,it has rich research significance to improve the efficiency of solving the above problems by mixing the advantages of different algorithms.A Hybrid Particle Swarm Optimization based on Multi Region Sampling strategy(HPSO-MRS)is proposed to solve the multi-objective FJSP.The main improvements are as follows: First,the Multi Region Sampling strategy(MRS)is used to reorganize the particles according to the position of the Pareto front,and plan the corresponding motion direction,so as to adjust the strong convergence ability of particles in multiple directions.Second,in the aspect of codec,the decoding strategy based on the interpolation mechanism is used to eliminate the local left shift in the decoding process,improve the quality of the decoding results and speed up the convergence process of the algorithm.Thirdly,the crossover and mutation operator of genetic algorithm is used to update the particle’s position,and the crossover operator based on exchange order is designed according to double vector coding,which improves the diversity of search process and avoids the algorithm falling into local optimum.In the experimental part,HPSO-MRS are compared with the other four algorithms.Experimental results show that the proposed algorithm has good convergence and distribution performance.Aiming at the multi-objective DFJSP,a Hybrid Particle Swarm Optimization based on Fast Multi Region Sampling strategy(HPSO-FMRS)is proposed.In addition to continue to use the previous effective improvement strategy,we mainly consider reducing the running time of the algorithm.The main improvements are as follows: First,use the Fast Multi Region Sampling strategy(FMRS).Which reduces the external archive scale,reduces the use times of the sorting algorithm,improves the efficiency,and ensures the diversity of the search process.Secondly,for the DFJSP of heterogeneous job-shop,a coding scheme with rich redundancy structure is designed to avoid illegal solutions.Thirdly,used the general probability parameters to update the vectors independently,so that the three decision vectors do not change at the same time,so as to ensure the global search ability of the algorithm.The experiments are compared with other five algorithms.Compared with the previous improved algorithm,this algorithm improves the efficiency,and the result quality is better than the original algorithm,which can achieve good convergence performance and distribution performance.Finally,the hybrid intelligent optimization algorithm proposed in this paper combine the ideas and characteristics of multiple algorithms,designs a multi region sampling strategy which can enhance the strong convergence ability of the algorithm in multiple directions of Pareto front.According to the characteristics of the problem,the corresponding optimized decoding and crossover operators are designed.In addition,the process of the improved method is further optimized in order to achieve better solution efficiency as well as better solution efficiency.The experimental results show that the proposed methods and strategys can better deal with multi-objective FJSP and multi-objective DFJSP,which provides a useful reference and ideas for hybrid intelligent optimization algorithm to solve other complex multi-objective optimization problems.