针对串行优化算法在搜索时间上的不足,提出了一类组合优化问题的并行粒子群算法。该算法将粒子群划分为多子种群异步并行运算,利用不同范围内的多极值,指导粒子速度更新,加入邻域搜索策略,提高了搜索速度,同时也有效地防止了粒子在最优点附近发生的振荡现象。仿真实验表明,该算法与其他搜索方法比较,在搜索时间和求解质量上具有优势。现已应用于钢铁生产热轧计划编制中,并用实际生产数据表明了该算法的可靠性。
A parallel particle swarm algorithm designed to solve a kind of combinatorial optimization problem was presented to overcome the heavy computational time disadvantage of general serial algorithm. The parallel algorithm performed asynchronously by dividing the whole particle swarm into several sub-swarms and updated the particle velocity with a variety of local optima. A local search strategy that prevented particle librating in the neighborhood of optimum was proposed. The parallel algorithm's validity was proved by a simulation test comparison with other algorithms. It was also applied to hot rolling planning, and a satisfactory result was achieved in production.