A Comparative Study of DWT and EEMD Methods for Validation and Correction of Pyroshock Data

Abstract

Low-frequency errors are a type of testing errors commonly existing in pyroshock experiments due to the harsh mechanical environment that lead to inaccuracy and contamination of pyroshock data. It is necessary to eliminate such low-frequency errors from pyroshock data via data analysis. Two methods based on discrete wavelet transform (DWT) and ensemble empirical mode decomposition (EEMD) are proposed, respectively, in this article. Firstly, the principle and operating process of these two methods are introduced, where four correlation coefficients are defined to determine the decomposition level for quantitative correction. Then, the two methods are applied to remove the integral zero shift from a set of pyroshock test data, respectively, and the capability and effectiveness of the two methods are compared. It is revealed from the comparison studies that both methods are capable of removing integral zero shift effectively to correct the shock response spectrum and improve the accuracy of pyroshock data. The correlation coefficients constructed here can help to realize quantitative selection of key parameters for the two methods. The DWT method shows better correction performance for extracting accurate parameters, whereas the EEMD method is more concise and convenient for application.