Detection of Mine Precursor Signals Based on Decision Tree Model

In order to prevent disasters more effectively, monitoring technology can capture different types of signals in electromagnetic radiation and acoustic emission signals in real time, and predict possible dangerous events based on the data. This article aims to detect the precursor signals that appear before the danger of the mine, providing technical support for coal mine safety detection. This article uses time and frequency domain dimensions to process precursor signals to obtain the mean, variance, kurtosis, pulse count, spectral energy, spectral components, spectral peak, and spectral density of precursor signals. In order to effectively identify the precursor signal interval of electromagnetic radiation and acoustic emission signals, sliding window feature extraction is used. We need to divide the data into windows consisting of 50 discrete data points for sliding, batch processing, and feature extraction. The decision tree model is used to train the precursor signal and non-precursor signal data, and the trained model is used for data prediction. Finally, the article achieved the prediction and identification of precursor signal models, thus realizing the role of disaster prevention.