一种综合改进型遗传算法及其在综合负荷建模中的应用

Synthetically improved genetic algorithm in power load modeling

针对基本遗传算法本身存在收敛速度慢和易早熟的缺陷,提出一种综合改进型遗传算法并成功地应用于负荷建模。该改进遗传算法通过对初始种群的选择、最优个体的保留、自适应的交叉和变异率、早熟现象的防止策略等各方面进行综合的科学设计,能十分有效地克服早熟、避免近亲繁殖、明显提高收敛速度,并具有优良的自适应特性。基于现场实测负荷特性数据的负荷建模实践表明,所提出的综合改进型遗传算法对于加速收敛缩短辨识时间、提高模型拟合精度、克服模型参数的分散性均具有显著作用,是一种很适合于负荷建模的优秀优化算法。

Aiming at the shortcoming of basic genetic algorithm as slow rapidity of convergence and easy to precocity, this paper presents a synthetically improved genetic algorithm and has applied it into power system aggregate load modeling. This improved genetic algorithm has comprehensive scientific designs in many aspects such as choosing original colony, reserving the best individual, adaptive crossover and mutation probability, preventive policy to precocious phenomenon, which can overcome precocity effectively, avoid close relative propagation, enhance rapidity of convergence obviously, and has good adaptive characteristic. The practical modeling based on the field measured data from power substation proves that this improved genetic algorithm has great effects on shortening convergence time, improving model precision, conquering the decentralization of parameters, and is an excellent optimum arithmetic and fairly fits for power system load modeling.