Discrimination of seismic signals using fuzzy entropy and a new FLVQ method

In this paper, a new clustering technique is introduced for seismic discrimination purposes, based on the P-wave spectra computed from short period teleseismic recordings. In this study, we have proposed an extended scheme of a fuzzy LVQ (learning vector quantization) model for clustering the six spectral features, extracted from the seismic signals. The model has been proposed as an extension of the FLVQ scheme presented by Sakulaba et al. (1991). The extended FLVQ model is combined with an additional sub-algorithm which uses the newly defined fuzzy entropy to find the optimum number of the clusters. The proposed clustering model has been tested on a set of 26 natural earthquake and 26 artificial explosion data. A comparison has also been made between the aforementioned clustering method with the conventional ones, using the leave one out testing strategy. Regarding the error rate, the experimental results are promising and some remarkable advantages of the newly proposed model are also discussed