Using AR Model and BP Neural Network to Identify Microseism Signal

According to the characteristics of broad frequency and abundant spectral components of mine microseismic signal, we use AR model parameters and BP neural network to propose a method of filtering treatment for the signal and noise with different frequency ranges. We can use this method to separate noise and signal, and decompose different frequency band signals, so we can achieve the goal of filtering. The experimental results suggest that we can effectively remove the noise of microseismic abnormal signal by using AR model parameters and BP neural network, and this method can be used in the microseismic prediction and the pretreatment of microseismic signal.