基于分层匹配追踪算法的电能质量复合扰动参数辨识方法

针对电能质量中的复合扰动信号分析问题,提出一种粒子群优化(PSO)和匹配追踪(MP)算法相结合的分层搜索的原子分解方法。首先应用MP算法提取基波分量,对于去除基波分量的残差信号,利用快速傅里叶变换找寻能量最大的频率成分,采用PSO算法粗搜索出最佳匹配粒子,然后以最佳匹配粒子为中心,在一定范围内重新离散化,生成小规模原子库,再应用MP算法有针对性地进行细搜索,最终得到最佳匹配原子,提取出电能质量复合扰动特征参数。仿真结果表明,该方法能克服MP算法匹配时间长、计算量大及PSO优化MP算法残差积累过大、容易陷入局部最优、匹配参数不准确等缺点,且具有一定的抗噪性和实时性。

Parameter identification based on hierarchical matching pursuit algorithm for complex power quality disturbance

An atomic decomposition method based on the hierarchical matching pursuit algorithm combining PSO(Particle Swarm Optimization) and MP(Matching Pursuit) is proposed for analyzing the complex power quality disturbance. MP algorithm is applied to extract the fundamental frequency component and FFT is then used to search the frequency component with the maximum energy. PSO algorithm is applied to extract the best matching particle from the residual signals and re-discretization around the particle within a certain range is then used to generate a small-scale atom library with the particle as its center. MP algorithm is applied again to purposely search the best matching atom for extracting the characteristic parameters of power quality disturbance. Simulative results show that, with a certain anti-noise capability and real-time performance, the proposed method avoids the defects of MP algorithm, such as long matching time and great computation load, as well as the defects of PSO-MP algorithm, such as excessive residual accumulation, easy local optimum and inaccurate matching parameter.