The Reserch On Multi-objective Production Scheduling For Color Coating Based On Intelligent Optimization

Color-coated steel plate is a kind of high value-added product which is deeply processed in iron and steel enterprises.In order to solve the problem of color coating production scheduling,most enterprises still adopt traditional methods such as manual experience scheduling and single objective scheduling.It is a typical multi-objective optimization problem that many conflicting optimization objectives and complex process constraints need to be satisfied simultaneously in color coating production scheduling,which makes it difficult for traditional methods to generate high-quality production schedules for color coating process.Therefore,this thesis conducts an in-depth study on the multi-objective production scheduling problem of color coating,establishes the multi-objective integer optimization model for the problem,and proposes an improved MOEA/D algorithm based on the new coding to efficiently solve the problem.Specific research contents are as follows:(1)According to the process of color coating production,the complex technological constraints in actual production are summarized,three optimization objectives that affect the unit productivity are extracted,and the conflicting relations among multiple objectives are analyzed.(2)According to the process constraints and characteristics in the process of color coating production,three optimization objectives including color switching number,inner diameter switching number and thickness switching number,as well as the constraints corresponding to the actual production process of color coating,are established,and a multi-objective integer optimization model of color coating production scheduling problem is established.(3)In view of the traditional coding method that is difficult to effectively solve the actual scheduling problem,this thesis proposes a segmentation encoding method,in which the constraints are embedded,and the corresponding evolution operator,and develops a discrete NSGA-II for the color coating scheduling problem in a practical steel plant.The optimization results are compared and analyzed with the simulation of traditional manual scheduling method.Experimental results show the validity of the proposed encoding method and the superiority of the multi-objective evolutionary algorithm in solving this problem.(4)In view of the deficiency of NSGA-II that is easy to fall into local optimization and poor search dispersion,this thesis embeds machine learning into MOEA/D and uses the idea of K-means clustering to develop an improved MOEA/D algorithm with local search(MOEA/D-CLS)for the problem.Experimental results based on actual data show that MOEA/D-CLS has higher convergence speed and search dispersion,which is obviously better than NSGA-II.