Research On Job-Shop Scheduling Based On Optimal Neural Network

With the rapid development of computer intellectualization,manufacturing enterprises are also following the pace of transformation towards intellectualization,in order to save manpower and material resources to a greater extent,improve production efficiency and maximize profits.This requires that we strive to improve the existing productivity by means of modern science and technology in all aspects of production.Job-shop scheduling,as an important intermediate link,plays an important role in the production capacity.Most of the existing research methods are about scheduling attributes mining and solving algorithms.In these two aspects,scholars have done more research on solving algorithms.Generally speaking,the main research focus of solving job-shop scheduling problem is the combination of multiple hybrid intelligent algorithms.Nowadays,different combinations of neural network and other algorithms are the mainstream.Simulated annealing and genetic algorithm are not as effective as particle swarm optimization in convergence speed.The advantages of training neural network based on particle swarm optimization have not been applied to job-shop scheduling in the existing research.Therefore,this topic proposes the neural network which will be optimized.The network is applied to job shop scheduling.The single objective problem under deterministic conditions and the multi-objective job-shop scheduling problem under stochastic uncertainties are studied.This paper mainly does the following work:(1)This paper combines particle swarm optimization algorithm with BP neural network algorithm,and solves the job-shop scheduling problem based on the optimized neural network,which aims at achieving the minimum completion time.A job-shop scheduling model based on neural network is established.The weight of the network is optimized by using particle swarm optimization algorithm to solve the defect of local optimum of the neural network.(2)The optimal RBF neural network and genetic algorithm are combined to solve the multi-objective job shop scheduling problem under stochastic uncertainties.Restricted by different production environments,job shop scheduling problems can be divided into different types,some are continuous deterministic scheduling problems,and some are random uncertain scheduling problems.In this paper,on the discrete uncertain job shop scheduling problem,referring to the research status of scholars,the stochastic programming method is used to establish the multi-objective function on the uncertain problem.Because of the complexity of the function model,it is necessary to use the neural network to approximate the function in the calculation of the multi-objective function.Finally,the genetic algorithm is used to solve the problem.Based on their framework,an improved particle swarm optimization(PSO)algorithm is introduced to train RBF neural network,which can accelerate the convergence speed and improve the efficiency of the algorithm.