Research on the Prediction of Rockburst Risk in Deep Coal Mining Based on Random Forest and Support Vector Machine Models

This article studies and develops a prediction model that combines random forest and support vector machine (SVM), aiming to solve the safety problems caused by rockburst during deep coal mining in China. This model accurately extracts precursor characteristic signals from acoustic emission (AE) and electromagnetic radiation (EMR) monitoring data, achieving accurate warning of rockburst. Data preprocessing includes using summary functions and K-S tests to ensure data integrity and distribution characteristics, as well as using box plots to handle outliers. The feature extraction stage focuses on extracting key information from the statistical and frequency characteristics of the signal, and using Fourier transform to analyze the signal. The identification of interference signals adopts a logistic regression model, and the random forest model is used to enhance the recognition ability of precursor characteristic signals. Support Vector Machine (SVM) processes nonlinear relationships and displays prediction results through visualization. Research has shown that the model is innovative in theory and demonstrates high accuracy and reliability in practical applications, providing a new technical approach for early warning of coal mine rockburst and leading the way in mining engineering safety management.