电力系统谐波分析的稳健支持向量机方法研究

随着电力系统谐波污染问题日益加剧,准确地掌握电网中谐波的频率分量对于电力系统的安全、经济运行具有重要的意义.文中采用基于支持向量机的稳健频谱估计算法用于电力系统谐波和间谐波的分析;采用迭代变权最小二乘法克服了常规算法中计算复杂度和时间序列的长度成指数性增长的困难;通过引入特殊的代价函数的方法消除异常值影响,使算法对异常值具有稳健性.该算法的优点是精度高,鲁棒性强,对异常值和脉冲性噪声不敏感,具有很强的稳健性.通过对受到脉冲噪声污染的直流电弧炉电流波形进行分析,同时与快速傅立叶变换和Prony方法进行对比分析,表明了该法在没有异常噪声的情况下和有大量异常噪声干扰的情况下都有相当高的分析精度,可以满足电力系统谐波和间谐波分析的要求;在matlab6.5环境下编程,计算速度也满足要求.该算法的不足之处在于为满足稳健性的要求,模型参数的选择需要模型的先验知识或由采样时间序列的统计量根据经验进行选择.

A ROBUST SUPPORT VECTOR ALGORITHM FOR HARMONICS ANALYSIS OF ELECTRIC POWER SYSTEM

With the deterioration of harmonics pollution in power system, it is of great importance to accurately find out the harmonics component for the safe and economical operation of the power system. A novel robust approach to harmonics and interharmonics analysis, based on Support Vector Machines and solved by Iterative Reweighted Least Squares algorithm to overcome the difficulty of exponential computation complexity, is proposed in the paper. By introducing specific loss function, the method can mitigate the infection of outliers and noises and exhibits robustness characteristics. The proposed method also has high analysis precision. The application to impulse noises polluted signal of dc arc furnace installation without compensation and performance comparison with fast Fourier transform (FFT) and Prony algorithm show that the proposed method has good harmonics analysis precision to the signals both in normal and much impulsive noises contaminated conditions. Run in the Matlab version 6.5 environment, computing speed satisfies application's requirement. The shortcoming of the algorithm is that its parameters should be tuned for the best performance according to prior knowledge or the statistics of the analyzed signal.