基于改进SCE算法的Job Shop调度方法研究

An Improved and Effective Shuffled Complex Evolution Algorithm for Job Shop Scheduling Problems

Job Shop Problem(JSP)是生产调度领域中的一类复杂的调度问题,高效JSP求解算法的研究与设计是JSP,乃至整个生产调度领域的关键研究内容。引入新的智能仿生算法Shuffled Complex Evolution(SCE),以求解工件的最小最大完成时间为目标,通过序列映射方式将连续定义域空间中的变量映射到离散的组合优化问题空间中,同时采用基于工序编码的方式进行编码,最后使用顺序插入解码机制对其解码。并针对基本SCE算法在求解优化问题时求解质量差和求解速度慢等缺点,对算法中个体的进化过程进行改进,使个体进化的方向沿着当前群体最优解的方向进行。最后将此算法用于求解典型的Job Shop调度实例,结果表明,改进SCE算法在解决Job Shop调度问题上是有效的。

The job shop problem is an important content of scheduling in the manufacturing industry. The design of the high efficiency algorithm for JSP is the key to the production scheduling in manufacturing factory. A new intelli- gent algorithm, named Shuffled Complex Evolution （SCE） algorithm, is proposed in this paper with the aim of get- ting the minimized makespan. The sequence mapping mechanism is used to change the variables in the continuous domain to discrete variables in the combinational optimization problem; the sequence, which is based on job per- mutation, is adopted for encoding mechanism and sequence insertion mechanism for decoding. Considering that the basic SCE algorithm has the drawbacks of poor solution and lower rate of convergence, we use a new strategy to change the individual＇ s evolution in the basic SCE algorithm. The strategy makes the new individual closer to best individual in the current population. The improved SCE algorithm was used to solve the typical job shop scheduling problems and the results and their analysis show preliminarily that the improved algorithm is effective for such problems.