The Research And Application Of Improved Adaptive Non-dominated Sorting Genetic Algorithm In Multi Objective Of Job Shop

Shop scheduling is one of the important factors of production efficiency in a manufacturing enterprise.With the development of economic globalization and the intensification of market competition,manufacturers must ensure efficient and stable operation to continuously improve the commercial status.Therefore,the realization of enterprise management and production of reasonable arrangements for enterprises is the key to maximize profits.Scheduling problems are known as one of the most difficult NP problems.As a result,the shop scheduling problems had been researching thoroughly by academia and engineering in recent years.A multi-objective optimization problem is common for solving complex actual scheduling problems.It is necessary to take into account the various constraints of the problem before making various important decisions.And it is likely that the objectives are conflicting and the purpose of the multi-objective optimization is to resolve conflicts to get relatively optimal solution.In recent years,a variety of optimization algorithms have been introduced in the analysis of multi-objective shop scheduling.Traditional multi-objective optimization methods have many defects,such as optimal objective weighting method is just a simple weighted sum for each target,but the units of each goal is often different,which gives bigger subjectivity and not-operational for the distribution of the weighted value.There are many improvements had been made according to the characteristics of various algorithms at home and abroad.Non-dominated Sorting Genetic Algorithm(the NSGA-?)has good effect on the fitness assignment and diversity of the solution after investigation,but it has inevitable defects.In this paper,NSGA-? algorithm was analyzed and improved to overcome the defects that NSGA-? algorithm had on crowded distance calculation and the elite selection method.Using cyclic crowded distance to calculate,and eliminate the minimum distance,finally ensure the evenness of species distribution;Then select part of the individuals from each non-dominated solution according to the strategy that add elite to keep improvement of elite selection;Finally ensure the next generation of individual number according to the selection strategy.In this paper,improved algorithm is applied to job-shop scheduling,and this paper propose a self-adaptive genetic operator which is automatically adjust base on the evolution algebra to avoid "premature" phenomenon,and accelerate the convergence speed of the algorithm.The local convergence judgment is introduced to prevent the local optimal solution from jumping out of the local optimal and improve the global solution of the optimal solution.The Java programming language is used to implement the research of job-shop scheduling problem on the basis of the algorithm in this paper,and the performance of the algorithm is verified through several classic examples.Finally,the shop scheduling system on the basis of algorithm in this paper is achieved through the simulation of actual shop scheduling of an enterprise.