摘要
在电力系统谐波源特性表征问题的研究中,传统算法在建模速度和精度方面很难同时满足。为解决上述难题,提出一种改进粒子群优化的最小二乘支持向量机谐波源建模方法。根据统计学原理建立谐波源的最小二乘向量机模型,然后利用改进粒子群算法对模型优化。以VSI-PWM调速系统为例,采用遗传算法和改进粒子群算法对最小二乘支持向量机谐波源模型优化对比。结果表明采用改进的粒子群算法对谐波源进行最小二乘向量机参数优化,在计算时间和表征精度方面都要显著优于遗传算法。新模型有效地解决了速度和精度之间的矛盾,在提高建模精度的同时大大减少了训练时间,是谐波源特性表征的有效方法。
In the study of feature representation of harmonic source, it is difficult to meet both speed and accuracy in modeling by traditional algrithms. To solve this issue, a least squares support vector machine model of the harmon- ic source is proposed based on improved particle swarm optimization. According to the principle of statistical, the mo- dle of harmonic source is set up based on least squares vector machine and optimized by using improved particle swarm algorithm. By taking a VSI - PWM speed control system as an example, the improved particle swarm algorithm is compared with the genetic algorithm to optimize the harmonic source of least squares support vector machine. The result shows that the improved particle swarm algorithm has a greater superior to optimize least squares vector machine parameters than genetic algorithm no matter in the calculation time or accuracy. This modle is an effective method for harmonic source feature representation, it can sovle the contradiction between speed and accuracy effectively and re- duce training time while improving the modeling accuracy greatly.
出处
《计算机仿真》
CSCD
北大核心
2015年第9期158-162,共5页
Computer Simulation
关键词
谐波分析
谐波源建模
粒子群优化算法
遗传算法
最小二乘支持向量机
Harmonic analysis
Harmonic source modeling
Particle swarm optimization （PSO）
Genetic algo-rithm（GA）
Least squares support vector machine（LS-SVM）
