Research On Capacity Planning Of Cold Rolling Multi Production Line Based On Multi-objective Differential Evolution Algorithm

The iron and steel industry is one of the pillar industries of national development.And the cold rolling process is the last step of the steel production process.The type and output of steel products produced by each unit in the cold rolling process every day are decided by the cold rolling capacity plan on the basis of the capacity tasks and maintenance plans of various products given at the beginning of the month.The capacity plan directly affects the output and efficiency of steel enterprises.This paper studies the production capacity planning of cold rolling,which is characterized by multiple processes,multiple products and network production line structure in the production capacity planning process.The capacity plan should meet the actual complex process constraints and ensure the continuous production of the unit to increase the capacity;In addition,it is necessary to reduce the switching times of different products in the production process to improve the production efficiency.Aiming at this problem,this paper established a multi-objective optimization model for capacity planning,designed a heuristic algorithm based on human experience and planning rules,and designed a multi-objective discrete differential evolutionary algorithm based on the characteristics of the problem,and developed a cold-rolling capacity planning optimization system to meet the actual production needs.The specific contents are as follows:(1)For cold rolling capacity planning problems,in order to maximize production line production plan period and minimize switching cost as the optimization goal,considering the unit capacity,production path,the inventory balance,continuous feeding,practical process constraints such as priority,a multi-objective mixed integer programming model is established,each working procedure to decision-making within each unit production product types every day.An efficient heuristic algorithm for capacity planning was designed based on human experience and the rule of planning,and was successfully applied to actual production.(2)A multi-objective discrete differential evolutionary algorithm was designed to solve the productivity planning problem studied in this paper because the optimization software could not directly and efficiently solve it.Based on the characteristics of the problem,three improved strategies are proposed to improve the performance of the algorithm.The second selection strategy of elite individuals is put forward,which searches for better individuals around elite individuals after selection operation.In this paper,we propose the re-clustering of external file sets based on k-means,and generate the non-dominant solution to the clustering center.The comparative test results based on the actual production data show that the improved algorithm proposed in this paper is better than the representative multi-objective optimization algorithm.(3)Based on the models and algorithms mentioned above,the production capacity planning system is designed and developed,and a database is established to facilitate operators to conduct statistical analysis and management of various information in different dimensions.Through the embedded heuristic algorithm and multi-objective differential evolutionary algorithm,the production plan of process and unit is arranged to achieve the effect of automatic compilation and improve the productivity and production efficiency at the same time.