解空间遗传算法在水电站厂内经济运行中的研究

Study on solution-generated genetic algorithm applied to in-plant economical operation of hydropower station

针对水电站厂内经济运行求解方法,提出一种解空间遗传算法。采用避开空蚀振动区、考虑负荷平衡约束和机组出力约束的解空间初始种群生成法,以避免适应度函数设计中的惩罚处理并保证适应度函数非负;运用解空间摄动变异算子,保证变异后的个体仍为可行解。以三峡水电站为例与传统遗传算法进行了比较,不同量级的负荷分配结果表明:解空间遗传算法能够避开空蚀振动区的影响,保证机组的正常运行。同时由于避免了在不可行解区域寻优,改进算法提高了搜索的效率。改进算法避免了适应度函数中惩罚的处理,保持了种群的多样性,为改进遗传算法在水电站厂内经济运行中的研究提供了一种思路。

A solution-generated genetic algorithm （SGGA） is proposed for in-plant economical operation of hydropower station. This algorithm adopts a new method for generating initial population that can realize a non- negative fitness function to avoid the cavitations-vibration mode by imposing load balance constraint and unit output constraint on initial population generation, rather than imposing penalties on fitness function design by traditional genetic algorithm （TGA）. And a new perturbation mutation operator is introduced to guarantee a feasible solution to the individual after mutation. This SGGA method of in-plant economical operation was applied to the Three Gorges hydropower station and it was compared with TGA. The solutions for several typical loads show that the new algorithm can ensure normal operation of the units by avoiding cavitations- vibration mode, and it produces excellent solutions under the same environment via avoiding searching in the infeasible solution area. This study also offers a new idea for improving the genetic algorithm in its application to in-plant economical operation of hydropower station.