In view of the lag of the current microseismic monitoring system in the imminent warning of rockburst, the earth sound monitoring system is applied to monitor rockburst and the time series ARIMA model is studied in data prediction. The advanced 4-channel ares-5/e geoacoustic monitoring system is installed and continuous data monitoring is carried out. First, the multi-channel geoacoustic monitoring data are cleaned to eliminate random and interference data, and then the data are tested for stationarity and data difference analysis. Then, the ARIMA data model is established and the correlation of the model is estimated. After identifying the model parameters, the validity of the model is tested, Through the verified model, the ground sound data is predicted and recognized. Through the multi-level extraction of the long-term data monitored by the geological sound equipment of a mine and the prediction of the geological sound data value in the later period, it is found that the time series ARIMA model has excellent prediction advantages in the prediction of medium and short-term geological sound data.