基于匹配追踪稀疏分解的电能质量扰动检测

根据基于冗余字典的匹配追踪(MP)信号分解思想,提出一种电能质量扰动信号的参数检测与特征波形提取方法。在MP算法的每次迭代中,首先采用快速傅里叶变换(FFT)搜索能量最大的频率成分,然后采用基于离散Gabor原子中心区域的简化内积计算方法获得扰动参数的估计值,并以该估计值作为初始解,采用BFGS算法做局部搜索,进而获得精确匹配参数,并结合基于递归的内积计算确定扰动的起止时刻,最后根据电力信号扰动波形特点,设计合成字典,确定与扰动成分最匹配的波形。对单一和混合电能质量扰动信号的分解实验表明,该方法可以实现扰动参数的快速精确检测,进而有效提取扰动特征波形,并具有较好的抗噪性能。

Power quality disturbance detection based on matchingpursuit sparse decomposition

According to the idea of matching pursuit (MP) signal decomposition based on redundant dictionary, a method is proposed to detect the parameters of power quality disturbances and extract the disturbance waveforms. During each iteration of MP algorithm, firstly, fast Fourier transform (FFT) is used to search the frequency component with maximum energy; then, a simplified inner product calculation method based on the central region of discrete Gabor atoms is adopted to obtain the estimations of the disturbance parameters; further, taking those estimations as the initial solution, the BFGS algorithm is adopted to perform local searching, then the accurate matching parameters are obtained; and the recursive inner product computation technique is used to determine the start and end time of the disturbances; finally, in the light of power quality disturbance waveform characteristics, the compound dictionary is designed to determine the final matching waveform that mostly matches the disturbance component. The decomposition experiments for single and combined power quality disturbances show that the proposed method can detect the disturbance parameters fast and accurately, extract the disturbance feature waveforms effectively, and also possesses good anti noise performance.