For the problem that the signal of the die forging press was affected by nonlinearity, instability, and strong background noise interference in harsh environments such as high pressure, high temperature, high speed, and high vibration during long-term operation, a multiscale smoothing preprocessing method based on wavelet transform and S-G filtering was proposed. Among them, the wavelet transform method decomposed the signal in multiscal to effectively capture the characteristics and dynamic changes of the signal, adjusted the wavelet basis function and S-G filter parameters to meet the characteristics of the multiscale signal and denoising requirements. The inverse discrete wavelet transform was used to reconstruct the signal to realize the integrity, continuity, and smoothness of the signal. The results show that this method can effectively eliminate noise at the multiscale level while retaining effective information, which is superior to traditional moving average method, S-G filtering, wavelet transform forced denoising processing method and threshold denoising method.