Research On Applying Modified Particle Swarm Optimization Method For Parallel Batch Processing Machines In Fuzzy Manufacturing System

We extend the problem of scheduling batch processing machines with non-identical job sizes into fuzzy manufacturing environment with two fuzzy factors of fuzzy processing times of batches and fuzzy intervals between batches. The fuzzy manufacturing environment is more practical, so seeking the effective methods to solve scheduling problems in fuzzy manufacturing environment is of important theoretical value and practical significance.This paper mainly discusses the application of the particle swarm optimization algorithm on batch scheduling problem, especially on parallel batch processing machines in fuzzy manufacturing environment. The main and pioneering works of this paper are as follows:(1) we summarized the classification of the batch scheduling problems and the research progress in each class, generalized the main features and research methods of batch scheduling problems as well as the common used performance indices. Based on the traditional model of minimizing makespan on parallel batch processing machines, we utilized triangular fuzzy number in order to characterize the fuzzy processing times of batches and fuzzy intervals between batches, then concluded the mathematical model of minimizing makespan on parallel batch processing machines in fuzzy manufacturing environment.(2)After summarizing the basic principles of particle swarm optimization and several common modifying methods including introducing inertia weight, constriction factor and reproductive operations, we focused on describing the thought of dynamic adjusting modified particle swarm optimization. The algorithm can dynamically adjust the limit position of each particle that distributes in the cycle formed by the best positions of the population and the particle, was adopted to escape the local minimum which is the defect of standard particle swarm optimization. In order to enhance local search capability and balance the overall and local search capability, we combined the DAMPSO with quadratic interpolation.(3) We applied the DAMPSO algorithm and the modified DAMPSO algorithm on batch processing problems with parallel batch processing machines in fuzzy manufacturing environment, proposed a coding method using priority value and Batch First Fit heuristic is used to transform the paths into feasible batches. By comparing the two algorithms with Best Fit Longest Processing Time, First Fit Longest Processing Time, Genetic Algorithm, Particle Swarm Optimization, the experimental results show that modified DAMPSO and DAMPSO can solve the problem more effectively and the predominance of this algorithm is even more conspicuous as the scale of problem becomes larger, besides, combining DAMPSO with quadratic interpolation outperforms the DAMPSO algorithm. Thus, the proposed algorithm has better search ability, which indicates that modified DAMPSO is feasible in solving batch scheduling problems with parallel batch processing machines.