Aiming at the problem that there is a lot of noise interference in the warhead explosion shock wave signal, which seriously affects the interpretation and calculation of shock wave overpressure peak value and positive pressure time, a new method based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) combined with principal component analysis (PCA) is proposed. The shock wave signal noise reduction method. In this method, CEEMDAN is used to decompose the original signal to obtain the intrinsic mode function (IMF) and residual components; then the IMF component matrix obtained by decomposing is reduced in dimension by PCA, converted into a fully orthogonal principal component matrix, and then repeated A new quadrature signal is formed to suppress the modal aliasing phenomenon, and the new quadrature signal is decomposed and reconstructed by CEEMDAN to obtain a denoised signal. Compared with Bessel low-pass filtering and CEEMDAN noise reduction method, model experiments and measured data analysis show that CEEMDAN-PCA combined noise reduction method can effectively remove high-frequency noise in the signal, retain low-frequency effective information, and obtain better accurate peak and overpressure times.