论文部分内容阅读
针对库群长期优化调度求解效率低下、实用性差等问题,提出了多核并行禁忌遗传算法。该算法利用遗传算法适应性强、运算简单等优势,采用多种群计算方式保持种群多样性,以提高算法的全局收敛能力;在子种群进化过程中应用单向环迁移拓扑模型进行信息交换,以增强算法收敛能力;结合子种群间求解独立的特点,引入多核并行计算策略,并使用禁忌搜索思想避免重复计算适应值,提高了算法的求解效率。红水河流域10座水电站的计算结果表明,该算法可以显著缩短计算时间,获得良好的调度方案,是求解库群长期优化调度的一种有效方法。
Aiming at the problems of low efficiency and poor practicability of library long-term optimization scheduling, a multi-core parallel taboo genetic algorithm is proposed. This algorithm uses the advantages of genetic algorithm, such as strong adaptability and simple operation, and maintains population diversity by using multi-population computing to improve the global convergence of the algorithm. One-way ring topology model is used to exchange information in the process of subpopulation evolution. And enhance the convergence ability of the algorithm. Combining with the independence of solving subpopulation, multi-core parallel computing strategy is introduced, and the algorithm of tabu search is used to avoid repeated calculation of fitness value and improve the efficiency of the algorithm. The calculation results of 10 hydropower stations in Hongshui River Basin show that this algorithm can shorten the calculation time and get a good scheduling solution, which is an effective method to solve long-term optimal scheduling of reservoir group.