Large Scale Flexible Job Shop Scheduling Opitimization Based On Distributed Evolutionary Algorithm

In recent years,with the rapid development of China's manufacturing industry,how to effectively enhance the competitiveness of enterprises has been an urgent problem for manufacturing enterprises.As one of the important optimization methods in manufacturing industry,flexible job shop scheduling(FJSP)has been widely concerned and achieved some results.However,with the development of manufacturing industry to large-scale manufacturing,the scale of flexible job shop scheduling problem has multiplied.There are some problems in solving large-scale flexible job shop scheduling(LSFJSP)with current method,such as slow optimization speed and poor optimization effect.In order to solve the problem of large-scale flexible job shop scheduling optimization,this paper improves the existing distributed evolutionary algorithm framework.The framework of dimension distributed evolutionary algorithm based on Bayesian grouping is proposed in this paper.Furthermore,combining with the idea of population distributed evolutionary algorithm,a framework of population-dimension hybrid distribution evolution is proposed.The main work of this paper is as follows:1LSFJSP problem modeling.To solve LSFJSP with distributed evolutionary algorithm,this paper describes and models the problem of large-scale flexible job shop scheduling optimization to minimize the maximum completion time and total energy consumption.2Research on LSFJSP based on population distributed evolutionary algorithm.According to the analysis of the current research situation of population distributed evolutionary algorithm framework,LSFJSP research method based on population distributed evolutionary algorithm is proposed and the superiority method is verified by comparative experiments.The experimental results show that using population distributed evolutionary algorithm to solve LSFJSP can achieve a better optimization acceleration.3Research on LSFJSP based on dimension distributed evolutionary algorithm.In order to deal with high dimension and complex solution space question of LSFJSP,this paper further uses the idea of "divide and conquer" to divide LSFJSP into several small-scale problems.At present,most of the existing segmentation methods are based on the idea of random,which is not suitable for the practical optimization problems with coupling relationship between variables.In this paper,we propose a dimension segmentation method based on Bayesian network for LSFJSP dimension segmentation.First obtain the Bayesian network to express the coupling relationship between variables by learning the existing solution individuals.According to the obtained Bayesian network,the problem is divided.Besides,this paper proposes a reference vector selection method based on Metropolis criterion.Finally,the superiority of grouping method based on Bayesian network and the generality and validity of dimension distributed evolutionary algorithm architecture based on Bayesian grouping are verified by experiments.4Research on LSFJSP method based on hybrid distributed evolutionary algorithm.In this paper,a hybrid distributed evolutionary algorithm framework based on Bayesian network grouping is proposed.On this basis,the superiority and generality of hybrid distributed evolutionary algorithm framework are verified by experiments.