多目标优化算法求解导叶关闭规律问题

Multi-objective Optimization Algorithm for Solving the Problem of Wicket-gate Closing Law

针对目前的单目标优化方法没有均衡考虑导叶关闭规律各目标分量的问题,综合考虑影响导叶关闭规律优劣的水击压力、机组转速和尾水管最大真空度这三个因素,将导叶关闭规律优化问题抽象为含有多个自变量和极值点的多目标优化问题,构建导叶关闭规律问题的多目标优化模型,并使用实数编码的遗传算法求解导叶关闭规律的多目标优化问题,遗传算法不依赖于问题的具体领域,具有很强的鲁棒性,实数编码的方式可以避免算法解码时的时间消耗,正交设计法生成的初始种群个体分布均匀,有助于维持种群的多样性。使用基于遗传算法的多目标优化方法求解导叶关闭规律问题,与遗传算法优化结果相对比,实验结果表示了本文所提算法的有效性。

Aiming at the problem of single objective optimization without considering the balance ofeach target component, considering the three factors--water hammer pressure, rotating speed and tailpipe maximum vacuum degree which have influence on the problem of wicket-gate closing law, theoptimization problem of wicket-gate closing law can be abstracted as a multi-objective optimizationproblem with multiple variables and multiple extrema optimization problem. For solving the problemof wicked-gate closing law, the multi-objective optimization model is established, and real numberencoding method is used. Using the multi objective optimization based on genetic algorithm to solvethe problem of wicked-gate closing law, genetic algorithm does not depend on the specific field ofproblems, which has strong robustness. Real encoding method can be avoided decoding timeconsumption, which can save time. The orthogonal design method is used to make the individualdistributing evenly, which is helpful to maintain the diversity of the population. Compared the resultsof multi-objective optimization with genetic algorithm’s, the experimental results show theeffectiveness of the proposed algorithm.