Research On Order Planning Problem Of Cold Rolling Production Line Based On Multi-Objective Discrete Differential Evolution Algorithm

The cold rolling is the last stage of the whole production process of iron and steel enterprises.The high efficiency and effectiveness of the cold rolling line determine whether companies can deliver finished products to customers with satisfied quality and quantity in time.This thesis studies multi-objective order planning problem in cold rolling,which takes the practical production in cold rolling as background.Based on the given order requirement quantity of each unit,the concerned problem is to determine the processing sequence of the orders in each unit.Aiming at this problem,a multi-objective mixed integer programming model is established;and multi-objective discrete differential evolution algorithm and multi-objective discrete differential evolution algorithm based on clustering algorithm are designed respectively.The cold rolling order planning scheduling system is developed based on the developed model and algorithms.The main contents are as follows:1)For the multi-objective order planning problem in cold rolling,a multi-objective mixed integer programming model is formulated,with consideration of the production restrictions in acid rolling and continues annealing units.In the model,two kinds of optimization objectives are considered,the first objective is to minimize the overall setup cost caused by the changeover of the steel type,the second objective is to minimize the overall time deviation.2)For the problem,an improved multi-objective discrete differential evolution(IMDE)algorithm is designed to solve it.To improve the performance,an adaptive adjustment strategy in the algorithm in the mutation operation is proposed,and four different individual strategies for different individual search periods to generate the individual are designed.IMDE is compared with the classical multi-objective difference algorithm under a certain scale of data to verify the effectiveness of the strategy.3)In order to further improve the performance of the algorithm,a multi-objective discrete differential evolution algorithm based on the population reconstruction strategy of Density-Based Spatial Clustering of Application with Noise(DBSCAN)is designed to target the convergence of the population in the algorithm search process.In iteration process,the population usually has the characteristics of convergence and aggregation.The discrete Hamming distance is used to represent the distance between individual vectors in the discrete space,and the differential population is clustered after a fixed number of generations.Based on these three metrics,this thesis compare this algorithm with three variants to verify the effectiveness of the clustering strategy;moreover,this thesis compare this algorithm with non-dominated sorting genetic algorithm-?(NSGAII)and multi-objective genetic local search algorithm(MOGLS)which are widely regarded as the state-of-the-art algorithms for multi-objective optimization problems,with the aim to verify the effectiveness of the proposed algorithm.4)Based on the above models and algorithms,an order planning scheduling system is developed,in which the corresponding database is established.In this system,relevant functions consisted of data manipulation,scheduling,manual adjustment,plan print,historical planning query and more can be realized.The development of the system as a whole has achieved the effect of supporting decision making,which in turn can significantly improve the production efficiency of enterprises.