基于混合编码改进遗传算法的无功优化

简单遗传算法用于无功优化时存在收敛速度慢、容易陷入局部最优等问题。针对无功优化控制变量既有连续量又有离散量的特点,提出整数和实数混合编码的改进遗传算法。该改进算法在进化前后期采用不同的选择方式;依据交叉位和变异位控制变量的类型确定相应的实型或整型的算术交叉、小变异遗传操作,并且交叉率和变异率随进化代数变化;在目标函数中选用按指数规律变化的越界罚系数。IEEE14、118节点系统的仿真计算结果表明,改进后算法在全局寻优能力和收敛速度方面优于简单遗传算法。

Reactive power optimization using an improved genetic algorithm based on hybrid-code

The simple genetic algorithm is constrained by its poor converging performance and readily leads to local optimization.In terms of characteristics of the control variables used in reactive power optimization,the paper presents an improved genetic approach,which is based on hybrid-coding of integer and real numbers.In the modified algorithm,different reproduction patterns are employed at different stages.The operations of arithmetic crossover and mutation are determined according to the types of their variables,and their probabilities vary with the evolution time.The penalty coefficient,applied to the objective function,meets the exponential variation law.Results of calculation and simulation for IEEE14-and IEEE118 bus systems show that,in the aspects of convergence speed and global solution,the improved genetic algorithm is better than the simple one.