关键词: 混合流水车间排产问题, 负荷平衡, 选择概率, 差分算法, 双种群自适应差分进化算法

Abstract: To solve the Hybrid Flow Shop with non-identical parallel machine Load Balancing Scheduling Problem (HFS-LBSP),a mathematical model was formulated.The weighted summation of processing time load balancing cost and total parallel machine waiting time was taken as the load balancing comprehensive evaluation index.A Double Population Self-adaptive Differential Evolution (DPSADE) algorithm was proposed for global optimization purpose.In DPSADE algorithm,a new bi-population construction methodology and cooperative evolution mechanism was devised for the usage of maintaining the population diversity and evolution energy,and a self-adaptive parameter adjusting strategy in terms of stop iterations was introduced to enhance the capacity in escaping from the local optimal and keeping evolution alive.To improve the searching efficiency of the algorithm,a new population initialization method based on load balancing selection probability was designed at the initialization stage.Two example of production scheduling problem for car manufacturing and steel smelting processing were simulated,and the results showed that the load balance evaluation index of DPSADE algorithm had a significant reduction compared with Genetic Algorithm(GA),Differential Evolution algorithm(DE) and Self-Adaptive Differential Evolution algorithm (SADE).

Key words: hybrid flow shop scheduling problem, load balancing, selection probability, differential evolution algorithm, double population self-adaptive differential evolution algorithm