时序数据去噪中的小波策略及评价指标

引入评价时序数据去噪效果的均方根变化量、互相关系数、信噪比及平滑度4项指标,分析了不同小波策略的去噪效果与指标间的关系。结果表明,不同于小波分解重构阶段,小波策略阶段的去噪均方根误差与4项指标存在正相关趋势,且归一化后的综合指标与去噪均方根误差归一化值存在0.95的正相关性,在实测时序数据分析中,可用归一化后的综合指标代替未知的去噪均方根误差来评价去噪效果,并以去噪残差分布作为去噪可靠性定性指标。

Wavelet Strategies and Evaluation Indicator in Time Series Data Denoising

We introduce the gap of mean square error variances,corre-lation coefficient,signal-nosie ratio and flatness to analyze the relationship between the denoising result of different wavelet strategies and indicator.The result shows that the RMS of wavelet denoising strategy is positively correlated with the trend of the four indicators and the correlation coefficient is 0.95,which is different from wavelet decomposition and reconstruction phase.We use normali-zed composite indicator to replace the unknownRMS to evaluate the denoising effect,and regard denoising residual distribution as reliability qualitative indicator.