基于波形匹配端点延拓法优化的经验模态分解算法在铁路继电器参数降噪上的应用

Application of Empirical Mode Decomposition Algorithm Based on Waveform Matching Endpoint Continuation Method in Noise Reduction of Railway Relay Parameters

针对铁路继电器参数的噪声问题,为提取其有效信息,该文建立一种基于改进波形匹配延拓法优化的经验模态分解(EMD)算法,利用自相关函数分离含噪信号,小波阈值去除噪声的混合降噪模型。首先根据继电器参数特点对波形匹配方法进行改进,重新定义匹配误差度公式,并引入匹配精度误差系数,采用改进波形匹配延拓法优化EMD分解过程产生的端点效应,得到有效的固有模态分量(IMF)和余项;然后求解其自相关函数,并根据自相关函数图像结合噪声信号特征分离出含噪分量;最后对含噪分量进行小波阈值去噪,去噪后与剩余分量和余项结合,得到重构后的参数序列。同时,提出利用结构相似性(SSIM)评价指标,结合信噪比(SNR)、方均误差(MSE)指数对模型可靠度评判。通过结果分析,并与EMD分解后重构和小波阈值去噪方法作对比,证明该模型可优化铁路继电器参数的降噪效果。

Aiming at the noise problem of railway relay parameters,this paper establishes a hybrid noise reduction model based on empirical mode decomposition algorithm optimized by the improved waveform matching continuation method to extract the effective information.This model uses autocorrelation function to separate noisy signals and wavelet threshold to remove noise.Firstly,according to the characteristics of relay parameters,the waveform matching method is improved.The matching error degree formula is redefined,and the matching accuracy error coefficient is introduced.The end effect produced by EMD decomposition process is optimized by the improved waveform matching continuation method,and the effective IMF component and residual term are obtained.Then,the autocorrelation function is solved,and the noise signal feature is separated according to the autocorrelation function image.Finally,the noisy component is denoised by wavelet threshold.After denoising,it is combined with the residual component and remainder to obtainthe reconstructed parameter sequence.Moreover,the SSIM index,SNR and MSE index are used to evaluate the reliability of the model.Compared with EMD reconstruction and the wavelet threshold de-noising method,the noise reduction effect of the model is proved.