A technique for earthquake and karst collapse recognition based on wavelet packet fractal and neural network

The focal mechanism and propagation path of natural earthquakes and karst collapse are different, so the frequency characteristics of their waveforms are different, too. The wavelet packet fractal method can effectively extract the natural earthquake and karst collapse waveform characteristics, and the radial basis function （RBF for short） neural network can well identify two kinds of events, therefore by using RBF neural network based on wavelet packet this paper takes the natural earthquake and karst collapse recorded by Guangxi Earthquake Networks Center in recent years as an example to try to identify two kinds of event waveforms. The results show that the recognition rate of natural earthquake and karst collapse event is up 89.5%, suggesting it is an effective method to identify natural earthquakes and karst collapse.