This paper proposes a noise suppression method based on an improved LMS adaptive filtering algorithm. Signal processing system software is designed to utilize the LMS adaptive filtering algorithm in a digital signal processor (DSP) to filter the acquired voltage signals. After noise reduction, the signals are converted into digital signals and then output as corresponding graphics or waveforms. The genetic algorithm is used to optimize the LMS variable step size parameters, addressing the contradiction between the convergence speed and steady-state error of traditional fixed step size algorithms, thereby enhancing signal processing capabilities. Research indicates that the signal level ranges for the low-frequency and high-frequency channels after LMS adaptive filtering decomposition are [-6.516, 6.731] and [-2.991, 1.925], respectively. When the step size factor is set to 1/500, the signal denoising accuracy is higher. When the threshold function is set to a soft function and the number of filtering decomposition layers is set to 4, only one singular value occurs, and the denoising effect is better.
