Simulation Model Reduction Method And Efficient Intelligent Scheduling Algorithm For Semiconductor Production Line

The scheduling problem in semiconductor production line is facing with difficulties like large-scale,time-consuming evaluation and uncertainty event.Under this circumstances,both the quality and efficiency of the scheduling algorithm are needed to be considered.Taking the semiconductor production line as a research background,a model reduction method and an efficient intelligent scheduling algorithm is studied in order to shorten the time required to obtain the optimal scheduling scheme.The main work of this thesis can be described as follows:1.To reduce computer execution time,a model simplification method based on selective clustering ensemble algorithm and theory of constraints is studied.In this algorithm,machine importance is identified accurately based on selective clustering ensemble algorithm.Then simulation model is reduced according to the theory of constraints.Finally,a model accuracy measurement mechanism is built and a self-correction structure is also established to meet the requirement of the reduced model under the uncertain environment.2.An efficient intelligent scheduling algorithm based on genetic algorithm and optimal computing budget allocation is studied in order to solve the problem of the stochastic unrelated parallel machine scheduling problem with the objective of minimizing makespan.This algorithm is divided into a fast search stage and an exact search stage.Genetic algorithm is selected as an intelligent search mechanism,and optimal computing budget allocation is employed to intelligently allocate the huge simulation replications brought by repeated sampling in order to improve search efficiency.Simulation results show that,the model reduced method and the efficient intelligent scheduling algorithm studied in this thesis can effectively solve the problem mentioned above and have a strong practical value.