摘要
由于电网负荷波动较大,系统随机运行性强,系统中非线性设备都会引起严重的电力信号噪声。传统的EMD去噪方法是结合能量极小值寻找噪声信号和有用信号的分界点,主要适应于信噪比较高的信号中,在信噪比较低时会出现误判。因此提出了一种适合电力信号的基于遗传算法的EMD去噪方法。新方法基于电力信号与噪声信号的不同频带分布,对含噪的电力信号进行EMD分解得到多组IMF分量,并在数学排列组合的启发下重构电力信号,然后将遗传算法运用到寻找最优IMF分量组合中。仿真结果表明,采用遗传算法的EMD电力去噪方法无论在信噪比还是均方误差方面均具有一定的改善和提高,保证了电力供电的可靠性,提高了电力信号的检测精度。
As the large disturbance of power system load and strong random system operation exist in power sys- tem, nonlinear equipment generates serious noise. Traditional EMD denoising methods utilize energy minimization to find the demarcation of point of the noisy signal and the useful signal. These methods are mainly used in signals with high signal to noise ratio (SNR), but low SNR may lead to misjudge. Thus, an EMD denoising method based on ge- netic algorithm is proposed for power signal denoising. The proposed method decomposes the noisy power signal into several intrinsic mode functions (IMFs) based on the different frequent band distribution of the signal and noise. Then the power signal is reconstructed by the inspiration of mathematical permutation and combination, and genetic algorithm is used to find the optimal combination of IMF component. Simulation results show that the proposed meth- od based on the genetic algorithm has obvious improvement on both SNR and mean square error, which can ensure the reliability of electricity supply and accuracy improvement in the detection of electric signal.
出处
《计算机仿真》
CSCD
北大核心
2014年第10期123-127,共5页
Computer Simulation
基金
四川省科技支撑计划项目(2012GZ0009)
四川省电力公司科技项目(12H1541)
关键词
经验模态分解
固有模态分量
遗传算法
信噪比
Empirical mode decomposition
Intrinsic mode function
Genetic algorithm
SNR