In view of the existing wavelet threshold denoising method that unknown parameters selection in existing functions depends on experience and brings about insufficient denoising or signal distortion after denoising

a particle swarm optimization (PSO) algorithm based on relative wavelet entropy (RWE) was proposed for adaptive optimization of unknown parameters in wavelet threshold function to achieve adaptive noise reduction of rolling bearing vibration signal. A wavelet threshold function with two unknown parameters was improved. The relative wavelet entropy was used as the fitness function to optimize the unknown parameters and the optimal wavelet threshold function was obtained. The vibration signal of rolling bearing acquired by simulation and experiment was analyzed. Results showed that the optimized wavelet denoising method can filter the noise from the denoised signal better

the signal waveform after denoising had better smoothness

the signal-to-noise ratio was 294% higher than the hard threshold

and retained more details of the original signal

presenting better denoising performance and practical value.