To enhance the signal-to-noise ratio of nuclear pulse signals,a novel threshold function processing method based on a wavelet algorithm is developed,effectively meeting the fundamental requirements for noise reduction in nuclear pulse signals.Moreover,an optimized variable step size adaptive non-stationary signal denoising algorithm called EVSL is proposed for nuclear pulse signal analysis.Through synchronous processing of analog nuclear signals and original measurement signals,comparative evaluation with existing algorithms demonstrates that the EVSL algorithm considers convergence speed and steady-state accuracy and significantly improves the signal-to-noise ratio while preserving practical useful information.Therefore,it serves as an effective approach to enhancing the quality of nuclear pulse signals.