Seismic exploration has gradually turned to deep formation and complex geological structures. Because the heterogeneity of the formation medium affects the propagation of reflected waves, the seismic data have the property of complex interference. It is difficult to achieve the requirements of high-precision image. The related methods of data processing have been widely researched and applied with the development of EMD algorithm. This algorithm has also improved CEEMDAN algorithm, which effectively solves the mode mixing problem of EMD. But CEEMDAN algorithm still has shortcoming in the ability to filter noise. The algorithm based on CEEMDAN (Complete Ensemble Empirical Mode Decomposition with Adaptive Noise) and permutation entropy is using by the seismic data having the property low signal-to-noise ratio. Firstly, the seismic signals were decomposed by CEEMDAN to obtain Intrinsic Mode Functions (IMF) from high frequency to low frequency components. Then, the IMF components are sorted using permutation entropy to retain useful information for reconstruction.Subsequently, the reconstructed signal is decomposed using CEEMDAN, and the IMF components suitable for reconstruction are selected by comparing parameters such as approximation degree, smoothness, and objective function value. This achieves effective suppression of complex interference signals by combining the permutation entropy algorithm. The results show that compared with the traditional EEMD algorithm, the CEEMDAN-PE algorithm has higher computational efficiency, not only solves the mode mixing problem, but also effectively protects the high-frequency effective signal of the reflected wave while suppressing noise. Moreover, it can effectively screen out abnormal signals by determining whether they exceed the permutation entropy value and whether the objective function value is closer to zero, ensure the accuracy and stability of the reconstructed signal, and provide an effective processing method for improving the signal-to-noise ratio of seismic signals.