Coal mine gas and coal dust explosion sound recognition method based on wavelet packet energy

At present, it is difficult to meet the emergency rescue needs of gas and coal dust explosion accidents due to the rate of missing and false alarms in coal mine gas and coal dust explosion monitoring. In order to solve the above problems, a coal mine gas and coal dust explosion sound recognition method based on wavelet packet energy is proposed. This method installs mine-used pickups in the key monitoring areas of the coal mine to collect the working sound and environmental sound of the coal mine equipment in real-time. The wavelet packet energy ratio of sound is extracted through wavelet packet decomposition, and the feature vector characterizing the sound signal is formed. The feature vector is input into the BP neural network to obtain the sound recognition model of coal mine gas and coal dust explosion. The wavelet packet energy ratio of the sound signal to be measured is extracted and input into the model as the feature vector to recognize the type of sound signal to be measured. According to the requirements of feature vectors and output results, a BP neural network with 8, 8 and 1 nodes in the input layer, hidden layer and output layer is established to train the recognition model. By analyzing the results of wavelet packet decomposition of underground acoustic signals in coal mines, it is confirmed that the Haar wavelet function is used and the number of wavelet packet decomposition layers is chosen to be 3. The experimental results show that the energy proportion of gas and coal dust explosion sound decomposed by wavelet packet is obviously different from other sounds. The wavelet packet energy proportion distribution of the same sound signal with different time is stable. Therefore, the wavelet packet energy proportion can effectively represent the features of the sound signal and has strong robustness. BP neural network training speed is fast, and only a small number of training iterations can achieve the expected error. The recognition accuracy is up to 95% in the presence of many disturbing sound signals in the coal mine. BP neural network has the best recognition effect compared with the extreme learning machine model and support vector machine model.