基于协同进化算法的西北电网无功优化

西北电网全网无功分布具有典型的结构性特征。而常规遗传算法忽略了无功优化问题的结构性特征,计算时间过长,甚至找不到最优可行解。就此,采用协同进化算法对西北电网的无功优化问题进行研究,将西北全网划分成了甘青、陕西、宁夏3个子区,建立了每个子区自身的目标函数,采用常规遗传算法对子区内的控制变量先行优化,并将运算结果作为下一步协同进化计算的初始值,从而预先对参与运算的初始点进行了有效地筛选,以提高运算效率。实际运行算例表明,在收敛速度和迭代精度上,此协同进化算法均优于常规遗传算法。

Northwest Grid Reactive Power Optimization Based on Cooperative Coevolutionary Approach

The reactive power distribution in Northwest Grid performes some typical structural characters. Generally genetic algorithms GA not only ignores these structral characters, but also costs too much calculating time. Even worse, GA may not be able to solve the optimization problem. According to the structural characters, Northwest Grid is partitioned to three divisions which are GansuQinghai division, Shaanxi division, and Ningxia division. Each division has its own objective function optimized with GA.The division's optimization results will be used as the initial variables for the coevolution. Through this method, the calculating efficiency will be raised by filtering the initial variables. It's demonstrated that, coevolution based on GA is better than GA, when comparing the convergence speed and iteration accuracy.