DEEP LEARNING-BASED MICRO-SEISMIC SIGNAL CLASSIFICATION AND IDENTIFICATION METHOD

摘要

A deep learning-based micro-seismic signal classification and identification method, comprising the following steps: step 1, establishing a sample database of micro-seismic signals and blasting signals; step 2, extracting features of dominant frequency, post-peak attenuation coefficient and energy centroid coefficient of sample signals, so as to form a sample feature data training set and a test set; step 3, using the sample feature data training set to train a deep neural network classification and identification model, using test set data to verify a classification and identification effect of a signal classification and identification model, and continuously improving the classification accuracy by means of cross training; and step 4, extracting a feature vector of a signal to be identified, and inputting same into the signal classification model to obtain an identification result. The method has the characteristics of simple algorithm, high adaptability and real-time performance, and high identification accuracy, and can effectively classify coal mine micro-seismic signals and blasting signals.