用于铁磁谐振过电压故障辨识的VMD参数优化方法研究

Research on Data Feature Extraction Method for Ferroresonance Over-voltage Fault Identification

准确辨识铁磁谐振故障可为启动消谐装置提供重要的依据,针对单相接地、弧光接地和铁磁谐振三种过电压故障频率混叠导致数据特征提取困难的问题,提出变分模态分解(Variational Mode Decomposition,VMD,VMD)的优化方法:首先分析各过电压故障信号的特征与区别,指出VMD参数对故障信号分解的影响;再利用NSGA-Ⅲ(Non-dominated Sorting Genetic AlgorithmⅢ,NSGA-Ⅲ)优化算法对VMD分量个数、惩罚因子等参数组合进行搜索,并依此确定变分模态分解算法的关键参数,利用参数优化变分模态分解算法对故障信号进行处理;最后,利用实际电网参数搭建过电压故障仿真模型,获得各种工况下的故障数据,并对故障信号进行提取,结果表明,利用NSGA-Ⅲ算法进行优化后的VMD在对故障信号进行特征提取,能够清晰的反映不同类型故障的特征,且对数据采样频率不敏感,与原有的VMD算法效果相比,能够更有效应用于各类故障的辨识.

Accurate identification of ferromagnetic resonance fault can provide an important basis for starting the harmonic elimination device.In view of the difficulty of data feature extraction caused by the aliasing of three overvoltage fault frequencies of single-phase grounding,arc grounding and ferromagnetic resonance,a variational mode decomposition is proposed(VMD)optimization method:firstly,analyze the characteristics and differences of overvoltage fault signals,and point out the influence of VMD parameters on fault signal decomposition;then use Non-dominated Sorting Genetic AlgorithmⅢ(NSGA-Ⅲ)optimization algorithm to search the number of VMD components,penalty factor and other parameter combinations,determine the key parameters of variational modal decomposition algorithm,and use parameter optimization variational modal decomposition algorithm to process fault signals Finally,the overvoltage fault simulation model is built by using the actual power grid parameters to obtain the fault data under various working conditions and extract the fault signal.The results show that the VMD optimized by NSGA-Ⅲalgorithm can clearly reflect the characteristics of different types of faults and is insensitive to the data sampling frequency,which is different from the original VMD algorithm Compared with the effect,it can be more effectively applied to the identification of various faults.