The introduction of advanced data processing and prediction models can effectively improve the accuracy and timeliness of coal mine safety supervision and reduce safety hazards. In this paper, an artificial intelligencebased coal mine gas monitoring and prediction method is proposed, and an improved LSTM-TimeGAN model is constructed by processing and analyzing the gas monitoring data. The method firstly utilizes the LSTM model to predict the environmental factors such as temperature and humidity, and then generates the gas data through the improved TimeGAN model and combines it with the LSTM to predict the gas concentration. The experimental results show that the prediction accuracy using this model is significantly better than the traditional method. Specifically, the prediction results using the improved LSTM-TimeGAN model are 0.01163, 0.06265, and 0.00476 in MSE, RMSE, and MAE metrics, respectively, which are significantly lower than those of the traditional TimeGAN and LSTMTimeGAN models. The model not only captures the time dependence of gas concentration, but also effectively improves the stability of data generation. With this method, more accurate early warning of gas concentration can be provided in actual coal mine production to effectively improve safety.
