Real time seismic event detector using neural networks

A real time Seismic Data Automatic Acquisition System was designed and implemented int he Venezuelan Andes Seismic Network, operated by the Los Andes University Geophysical Laboratory. The system works on a PC with additional A/D data acquisition card and a driver, both of which were especially designed to detect seismic events. In order to diminish false event detection, neural networks are used to recognize seismic patterns in the frequency domain, after applying a time domain filter algorithm of the STA/LTA (Short Term Average to Long Term Average) type, as commonly used in many systems. The developed algorithm is so efficient that it can operate ion real time in practically any personal computer. There is an additional advantage: seismic event detection is independent on the seismic signal noise level. It depends only on the knowledge base patterns. The system has been opening in the Venezuelan Andes Seismic Network since 1996 without any loss of information.