Research On Production Scheduling Of Reentrant Lines Based On Intelligent Optimization Algorithm

Reentrant lines are a class of complex production systems abstracted fromsemiconductor and film manufacturing systems. The basic characteristic of thereentrant lines is that a job visits certain machines more than once. With thedevelopment of microelectronics industry, reentrant lines have drawn much attentionof academia and industry.The research of scheduling reentrant lines not only has hightheory value but also high economic value. This paper will study the reentrantscheduling problem and mainly discuss the following three aspects.Firstly, we study the reentrant scheduling problem on parallel machines with aremote server. We design a hybrid tabu search and a hybrid genetic algorithm forsolving this problem with minimizing the makespan. As only one start-point is used inthe traditional tabu search algorithm, a Restart method is established in the hybridtabu search algorithm. A heuristic algorithm is used to generate some individuals ofinitial population in the hybrid genetic algorithm. The numeric experiment shows thatboth the hybrid tabu search and the hybrid genetic algorithm perform well. Moreover,the hybrid genetic algorithm outperforms the hybrid tabu search not only in quality ofsolutions but also in the convergence speed.Secondly, the scheduling problem of reentrant job shop is discussed, a hybridshifting bottleneck algorithm is designed to solve this problem. The objective is tominimize the total weighted tardiness. The hybrid algorithm can be divided into twoparts, namely the sub-problem which is solved by the sequencing algorithm(SAL)based on the apparent tardiness cost (ATC) rule, and the rescheduling stage which issolved by the tabu search algorithm. Experiment results present that the proposedalgorithm is effective.Finally, we study the scheduling problem of reentrant flow shop with limit buffercapacity. A hybrid genetic algorithm is proposed, and the Nawaz-Enscore-Ham (NEH)algorithm is combined into the genetic algorithm to provide one initial solution.Experiments results show the proposed algorithm has better performance than otherstested.