REAL TIME SEISMIC SIGNAL PROCESSING USING THE ARMA MODEL COEFFICIENTS AND A LVQ CLUSTERING MODEL

In this paper we have presented a new method for seismic signal analysis, based on the ARMA modeling and the LVQ clustering method. The study is based on the background noise of the seismograms. The ARMA model has been chosen and the model coefficients are then derived for the consecutive overlapped windows, A base model is generated by clustering the calculated model parameters, using the LVQ method. The time windows which do not take part in model learning, are named as the test windows. The model .coefficients of the test windows are compared to the base model coefficients through some pre-defined composition rules, and a normalized value is generated as a measure of similarity. The set of the consecutive normalized values generate a curve versus time windows indices called as the characterislic curves. The results have shown that the characteristic curve contains many vital information and can be used for seismological purposes.