Investigation on Comprehensive Evaluation Index of Wavelet Threshold Denoising Based on Combination Weighting Method

The traditional wavelet denoising effect evaluation indexes，such as the root mean square error，signal-to-noise ratio (SNR)，correlation coefficient，and smoothness，are suitable for known pure signals. However，the pure state of a monitored signal is often unknown in practical applications. To solve the limitations of traditional evaluation indexes，a comprehensive evaluation index based on the combination weighting method is proposed in this study. First，on the basis of theoretical analysis and case verification，the index component was determined as the root mean square error and smoothness from the perspectives of increase-decrease characteristics and physical meaning. Then，the threshold maximum method was used to normalize the two indicators. Finally，the entropy weight method and variation coefficient method were applied to calculate the corresponding weights of the normalized values of root mean square error and smoothness，and the linear combination was added. The decomposition scale and wavelet basis function corresponding to the minimum value of the comprehensive evaluation index were the optimal wavelet parameters. The analysis of three SNR (5，12.5，and 20 dB) simulation signals shows that the optimal decomposition scale and wavelet basis function are identified，and the noise reduction effect is optimal. Compared with other comprehensive indicators，the accuracy and universality of the index are better. In addition，this index can be applied to GNSS monitoring signals，which has certain engineering value. © 2022 Tianjin University. All rights reserved.