Identification of vehicle suspension shock absorber rattle noise based on wavelet packet feature fusion and GWO-LSTM

With the advancement of pure electric vehicles, the issue of rattle noise in suspension shock absorbers has increasingly become a critical factor affecting vehicle comfort. This paper proposes a method for rattle noise recognition based on wavelet packet feature fusion and the grey wolf optimizer-long short-term memory (GWO-LSTM) model, aimed at improving the accuracy and efficiency of rattle noise detection. The vibration signals of the shock absorbers are decomposed by wavelet packet decomposition (WPD), followed by extraction of wavelet packet energy (WPE) and wavelet packet fuzzy entropy (WPFE) features and feature fusion. Subsequently, the GWO algorithm is employed to optimize the hyperparameters of the LSTM model, enhancing classification performance. The results demonstrate that, compared to traditional methods, the GWO-LSTM model significantly improves classification accuracy and training efficiency, achieving an accuracy rate of 97.85%, particularly excelling in the recognition of both slight and serious rattle noise. This study provides an efficient and reliable solution for the automated evaluation of shock absorbers’ rattle noise.