tailieunhanh - Future Manufacturing Systems Part 14

Tham khảo tài liệu 'future manufacturing systems part 14', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | Hybrid particle swarm algorithm for job shop scheduling problems 253 convergence theory. This algorithm has made full use of the large scale random search capability and the social cooperation of PSO algorithm at the same time the local parallel TS algorithm is embedded to improve the local search capability. We apply the above convergence theory to computational experiment and find the optimum of problem FT10 LA02 and LA19 in a short period. When compared with PGA algorithm and PSO algorithm there are 10 average value in ten times experiments of HPSO algorithm better than PGA algorithm and the deviation between the average value of ten times experiments of HPSO algorithm and the optimum is lower than that of between PSO algorithm and the optimum by . Thus the overall searching capability of the algorithm is improved which has demonstrated the effectiveness of solving Job Shop Scheduling problem by HPSO algorithm. Summary The theorem of No Free lunch NFL shows that there isn t a single solution that adapts to all problems effectively. Therefore a hybrid algorithm of particle swarm optimization and tabu search algorithm TS for solving JSSP is proposed motivated by the strong global search capability of PSO algorithm and the good local search capability of TS algorithm. Meanwhile the convergence of HPSO is proved and experimental simulation results are given. 4. Strategies for Deadlock Elimination for JSSP using PSO Deadlock is a state that the requests of scheduling transactions which contest resources one another can not be satisfied. Namely the stagnancy is among transactions waiting for one another appears. When using a PSO algorithm to solve constrained optimization problems deadlock is one of the key problems need to be solved. In PSO algorithm we optimize the various operations of jobs based on PSO code. Because the different operations of the same job are studied as the separate object. Different objects can be in different machine queues. So this .