One-Class Support Vector Machine with Particle Swarm Optimization for Geo-Acoustic Anomaly Detection

Without prediction and prior warning, earthquakes can cause massive damage to human society. The earthquake research has been exploring, and researchers discover that earthquakes happen with many natural phenomena, earthquake precursors. Geo-acoustic signals may contain a good precursor signal to a seismic event. The Acoustic Electromagnetic to AI (AETA) system, a high-density multi-component seismic monitoring system, is deployed to record geo-acoustic signals across 0.1Hz 10kHz. This paper aims to detect the anomalies of geoacoustic signals that may contain earthquake precursors. This study employs the One-Class Support Vector Machine(OCSVM) to detect the anomalies and applies Particle Swarm Optimization (PSO) to optimize the parameters of OCSVM. The experimental results show that the proposed method obtains promising results concerning the abnormal detection in geo-acoustic signals of the AETA system.