负荷模型参数辨识的粒子群优化法及其与基因算法比较

粒子群优化法(PS算法)具有全局性能好、搜索效率高等优点。文中应用该算法进行电力系统负荷模型的参数辨识，并将其与模拟进化算法进行比较，发现PS算法在计算时间、全局性方面均有比较明显的优势。讨论了PS算法中用以调节全局搜索和局部搜索关系的权重ω与搜索效率之间的关系，并给出了适用于电力系统负荷参数辨识的ω值。提出了一种利用PS算法的收敛快速性来提高全局性能的工程实用方法，并对工程实例进行辨识，收到了良好效果。

PS ALGORITHM IN LOAD PARAMETER IDENTIFICATION AND ITS COMPARISON WITH GENETIC ALGORITHM

Particle swarm (PS) optimization is a computational technique. It has roots in artificial life and social psychology as well as in engineering and computer science. This paper introduced PS algorithm, which is quite immune to local optima and is fairly efficient in solving problems with complex hyperspace into the field of electrical load parameter identification. This application involves a suitable neighborhood distribution that assures the better global searching ability of PS algorithm. The convergent efficiency and searching ability of PS algorithm, genetic algorithm (GA) and evolutionary strategy(ES) are compared. That leads to the conclusion that PS algorithm is more efficient than GA and ES in load parameter identification. The effect of an important parameter w on deciding the searching ability in PS algorithm is discussed and the best w that fits for the problem of load parameter identification is presented. The method of parameter limits definition and its effect in the algorithm is also discussed. The expected result is obtained when PS algorithm with optimal w is applied to the field data.