论文部分内容阅读
针对目前进化计算求解并行机动态调度存在局部搜索能力不足、计算周期长的问题,引入问题分解思想和估计评价策略,提出一种基于差分进化(DE)算法与代理模型相融合的快速求解方法。采用基于机器编码的DE算法对上层设备选择问题进行粗搜索。分析下层单机问题关键性特征构建能够预测调度性能指标优劣的代理模型,利用估计近似值取代费时的精确求解,降低繁冗评价过程带来的计算代价。在最佳分配方案的指导下,基于工件编码和多变异策略的DE算法确定设备上工件加工的前后顺序,实现设备分配与工件排序两个决策层同步优化。通过仿真实验表明,该方法优于传统并行机求解方法,尤其对大规模的并行机问题体现了更好的求解质量。
In order to solve the problem of lack of local search ability and long calculation period in the dynamic computation of parallel computing for evolutionary computation, this paper introduces the idea of problem decomposition and the evaluation evaluation method, and proposes a rapid solution method based on differential evolution (DE) algorithm and agent model. Using the DE algorithm based on machine code to search the upper equipment selection problem. Analyze the key features of the lower stand-alone problem. Establish an agent model that can predict the quality of the scheduling performance index, and replace the cost-accurate solution with the approximate value of estimation to reduce the computational cost caused by the redundant evaluation process. Guided by the best allocation scheme, the DE algorithm based on the coding of the workpiece and the multi-variant strategy determines the order of the workpiece processing on the device, so that the device allocation and the workpiece sorting are simultaneously optimized. The simulation results show that this method is superior to the traditional parallel machine method, especially for solving large-scale parallel machine problems.