Research on Noise Reduction of Engine Vibration Signal Based on RLMD- Improved Wavelet Threshold Algorithm

Due to the high speed of vehicle travel and road bumps, the vibration signals of engines during operation are characterized by non-stationarity and low signal-to-noise ratios, making it challenging to diagnose faults based on signal feature extraction. A signal denoising algorithm utilizing robust local mean decomposition (RLMD) and improved wavelet thresholding has been designed. The signal is decomposed into several product functions using RLMD. Then, improved wavelet thresholding (IWT) is applied to each product function. Finally, the denoised signal is obtained by reconstructing these functions. The IWT function reduces the shortcomings of traditional methods through smooth transitions, better preserving signal characteristics. Analysis is done through simulation experiments and actual diesel engine signal measurements. This paper compares the proposed algorithm with the traditional empirical mode decomposition (EMD) and wavelet threshold (WT) method, showing significant improvements in denoising algorithm. This method can better balance noise removal and signal preservation when processing complex non-stationary signals, offering strong practical application value.