面向随机加工时间的车间作业调度

Research on Stochastic Processing Time Oriented Job-shop Scheduling

提出了面向随机加工时间的车间作业调度方法,认为在整个遗传进化过程中出现频率越高的个体对环境的适应能力越强,该个体对应的调度方案为较优方案,构造了用于解决加工时间为服从正态分布的随机变量的车间作业调度问题的扩展遗传算法。在算法中设计了考虑设备能力空间的解码算法以产生活动调度方案;在交叉/变异过程中通过设计的基因调整算法确保新个体的合法性,以满足工序约束;采用基于适应值的轮盘赌的选择策略控制遗传进化的方向,使算法快速收敛到最优解。仿真实验验证了该算法在企业实际随机车间作业调度中的有效性。

A job--shop scheduling approach was studied where the processing time was treated as stochastic variables. The individual with highest frequency was supposed through all generations, and presented the best solution in terms of the fitness function value. Based on this hypothesis, an improved GA to stochastic job- shop scheduling problems （SJSSP） was proposed. A decoding algorithm considering the capacity space of machines was proposed to obtain active- scheduling solutions. Crossover/mutation gene recombining algorithms were introduced to ensure the new individual to meet precedence constraints. The Roulette strategy according to fitness function values was adopted to control the evolution direction of the generation. In addition, an instance shows that the algorithm possesses great validity.