基于蚁群优化算法的批调度问题分析-analysis of batch scheduling problem based on ant colony optimization algorithm.docxVIP

ABSTRACTBatch processing machine (BPM) is one of very important model production scheduling problems. Theoretically, it is a breakthrough and an extension of classic scheduling theory. The BPM problem ignores the restriction on the number of jobs that can be processed on a machine simultaneously, which shows a new research direction in production scheduling. BPM, extracted from the last phrase of semi-conductor production, is a new kind of scheduling problem and has a very strong background in application. Meanwhile, BPM is NP-hard which shows that it is very important to design a simple and effective algorithm for solving the problem.Ant colony optimization (ACO) proposed recently is an effective meta-heuristic algorithm in solving combinatorial optimization problems. It is a constructive based meta-heuristic, which means a complete solution was constructed by adding solution part into it. Therefore, the ACO algorithm is suitable to solve BPM because of its characteristic in batching. In addition, the ACO algorithm is population-based meta-heuristic which has memory storage capacity. The combination of these characteristics makes the ACO algorithm as a very unique meta-heuristic for solving the BPM problem. However, ACO has some shortcomings in long run time and easily trapped in local optimal in practice. Therefore, base on the structure analysis of the BPM problem, several improving scheme was proposed to promote the quality and efficiency of ACO in solving BPM.The research focuses on the development of efficient algorithms for solving BPM problem. Ant system algorithms for solving different types of BPM problems were proposed. Specifically, the main research and innovations include:The application of algorithm ant system (AS) was studied in the single batch processing machine with non-identical job sizes. First, the mathematic model of the problem was proposed. Lower bounds and heuristics of the problem already existed were also presented. Two ant system