Time Series and Non-Time Series Models of Earthquake Prediction Based on AETA Data: 16-Week Real Case Study
Chenyang Wang, Chaorun Li, Shanshan Yong, Xin’an Wang, Chao Yang
Abstract
The Key Laboratory of Integrated Microsystems (IMS) of Peking University Shenzhen Graduate School has deployed a self-developed acoustic and electromagnetics to artificial intelligence (AETA) system on a large scale and at a high density in China to comprehensively monitor and collect the precursor anomaly signals that occur before earthquakes for seismic prediction. This paper constructs several classic time series and non-time series prediction models for comparison and analysis in order to find the most suitable earthquake-prediction model among these models. The long short-term memory (LSTM) neural network, which gains the best results in earthquake prediction based on AETA data extracted from the precursor anomaly signals, is selected for real-earthquake prediction for 16 consecutive weeks.