International Journal of Advances in Computer Science and Its Applications
Author(s) : M. J. E. SALAMI , R. AKMELIAWATI , W. ASTUTI , W. SEDIONO
The ability of the signal processing techniques to predict earthquakes may help to reduce the catastrophic effect of the earthquake. The earth\\\'s electric field signal is one of the features that can be used to predict the earthquakes (EQs) by analyzing the changes in its characteristic prior the earthquake. The signal is extracted using extended Linear Predictive Coding (LPC). This approach is based on the projection of the excitation signal on the right eigenvectors impulse response of the LPC filter. The resulting projected value is weighted by corresponding singular value, leading to an approximate sum of exponentially damped sinusoids (EDS). The extracted vector is used as input of the prediction system in order to determine the location of the incoming earthquake. Support vector machines (SVMs) method is applied as classification technique. The basic idea of SVMs is mapping non-linear training data into higher-dimensional feature space through the kernel function. This paper presents a detailed analysis of the earth’s electric field signal due to earthquakes which occurred in Greece. The earthquake occurred during 2008 is used as model to train the system and some of the earthquakes that happened between 2003 and 2010 are used to evaluate the performance of the proposed system. The result shows good accuracy of 96.67% for training phase and 77.8% for the testing phase.