Abstract: To realize noise suppression in seismic exploration, this paper constructs a convolutional neural network model suitable for classifying and identifying seismic wavelets. First, the activation function, convolution kernel size, and normalization layer of the convolutional neural network model are designed. Then, the constructed convolutional neural network is used to extract features from the time-frequency spectrum of the seismic signal. Classification and identification of noisy seismic signals. Experimental results show that the model has a high classification and recognition rate, better anti-interference ability and practicality, which lays a foundation for future work of seismic exploration.