梯级水库群优化调度并行自适应混沌整体退火遗传算法

Parallel Self-adaptive Chaos Whole Annealing Genetic Algorithm for Optimal Operation of Cascaded Reservoirs

为了充分利用当今流行的多核配置资源提高梯级水库群优化调度的精细化管理水平,提出梯级水库群优化调度并行自适应混沌整体退火遗传算法。该方法利用Fork/Join多核并行框架的分治法将自适应混沌整体退火遗传算法的初始种群递归分解为多个规模更小的子种群,并平均分配到不同的内核中同时进行优化计算,实现算法的粗粒度并行求解。红水河梯级水库群发电优化调度实例结果表明,该方法比串行计算能大幅度缩减优化求解时间,提高计算效率,而且其并行化设计模式可为其他群体智能算法的并行化提供参考和借鉴。

For raising the level of the fine demand in management of optimal operation of cascaded reservoirs,a parallel self-adaptive chaos whole annealing genetic algorithm( PSCWAGA) was proposed by taking full advantage of widespread multi-core configuration.The algorithm utilized the divide-and-conquer method in Fork / Join multi-core parallel framework to divide the initial population of self-adaptive chaos whole annealing genetic algorithm( SCWAGA) into some smaller subpopulations,which were parallelized simultaneously in different cores.With generation operation of cascaded reservoirs located on Hongshui River as a case study,the result shows that the algorithm can cut down the execution time and improve the computing efficiency.Moreover,the proposed parallel design can provide guidance for other swarm intelligence methods.