采用优化经验模态分解的电力谐波辨识方法

The method for harmonic identification based on optimal empirical mode decomposition

电力谐波的准确辨识对智能用电具有重要的研究价值和意义。针对基于屏蔽信号的经验模态分解(M-EMD)在谐波辨识中幅值误差较大、模态分解不完整以及屏蔽信号构建参数依赖经验值等问题,提出对待分析信号进行滤波和模态预提取,并采用协同混沌粒子群优化算法(CCPSO)对屏蔽信号的构建参数进行寻优。电力谐波仿真辨识实验证明,与M-EMD算法相比,文中所述的IM-EMD算法在谐波辨识的准确度和可靠性上有了明显提高。

Harmonic analysis is very important to smart electricity.This paper proposes a hybrid algorithm to deal with the problems of empirical mode decomposition based masking signal（M-EMD） used in harmonic analysis,such as huge amplitude error,imperfect decomposition and the dependence on empirical values in building masking signal.The proposed method applies the method of chaos cooperation particle swarm optimization（CCPSO） to search proper parameters for masking signals,and employs filter and frequency pre-extract to the original signal.The synthetic experiments demonstrate that the method proposed in this paper gets better accuracy and reliability.