The hierarchical decomposition parallel optimizing technique for large-scale production lines

Abstract

The scheduling problems of production lines like buffer allocation problems (BAP) are usually NP-Hard problems and are complex to deal with using traditional techniques, especially for large-scale production lines disturbed by such stochastic events as the breakdown of machines. For large-scale production lines consisting of unreliable machines, the hierarchical decomposition parallel optimizing technique (HDPOT) is proposed here. The technique combines with the aggregation method of the hybrid production line to decompose the original production line into 1 n  subsystems by following some principles. Moreover, the asymptotic optimum solution of the original production line is gotten by searching parallel the optimum solutions of the subsystems. Meanwhile, an improved simulated annealing algorithm is proposed for the parallel optimization of subsystems. The HDPOT finally is tested by an example of buffer allocation problems (BAP) and specified by numerical experiments. The results show that this technique can solve scheduling problems more accurately, especially for large-scale production lines.