An earthquake prediction model based on precursor window detection

Abstract

Abstract The research of the earthquake precursor signal anomaly is one of the main research directions of short-term and imminent earthquake prediction. An earthquake prediction method based on time precursory window is proposed in this paper, which is based on the low-frequency electromagnetic signals collected by AETA. Firstly, the prediction model of historical low-frequency electromagnetic signals is constructed by machine learning method. The model is used to detect whether the current time period is in the window of earthquake precursors. Furthermore, two algorithms based on single-site and group-site position prediction is proposed in this paper. The algorithm filters three or more stations within the effective distance range, and uses the probability of earthquake occurrence as the weight to locate the earthquake center, so as to predict the position of earthquake occurrence. Finally, the real data set is tested on the earthquake of Qingchuan County, Guangyuan City, Sichuan Province, on February 18, 2018. The experimental results show that the proposed model has a good prediction effect.