一种改进遗传算法及其在负荷建模中的应用

在基于量测的电力负荷建模中,模型参数辨识结果的分散性是困扰模型应用的难题,其解决途径之一是在确定合理的模型结构和进行合理的时变性综合前提下,采用具有全局寻优特性的优化算法。遗传算法是一种具有全局寻优能力的随机搜索算法,但其本身存在收敛速度慢和易早熟的缺陷。为此,提出一种改进的遗传算法,它包括精英策略,交叉策略和移民策略,具有克服早熟、避免近亲繁殖和自适应的优良特性。建模实例表明,提出的改进遗传算法在精度、参数稳定性和计算时间上都远远优于基本遗传算法。

An Improved Genetic Algorithm and Its Application to Load Modeling

In measured data based power load modeling,unsteadiness of model parameters identified is an obstacle in the application of the load model.One of the settlements is to use global optimization algorithm with reasonable model structure and reasonable time-varying change synthesis.GA is a randomoptimization algorithm with global optimum capability,but it has the disadvantages of slow convergence and precocity.To solve above disadvantages,an improved GA is proposed,which includes elite strategy,crossover strategy and immigrant strategy.This adaptive improved GA can avoid precocity and inbreeding.Load modeling instances prove that the proposed algorithm is superior to usual genetic algorithms in terms of precision,parameter steadiness and convergence speed.