An Early Warning System to Predict Earthquakes based on WT-ARIMA-ANN Model

A devastating earthquake is one of the Earth’s most destructive natural disasters. Real-time seismic monitoring facilities can evolve into Earthquake Early Warning (EEW) systems with the help of modern technology, property damage, reducing the number of casualties, and economic losses. The proposed EEW system’s goal is to supply a precise damage assessment just before an earthquake hits a certain area. The proposed method involves preprocessing, feature extraction, and model training. It is recommended to preprocess the data by first filtering it and then reformatting it. Filtering is essential for getting rid of irrelevant or noisy information. In feature extraction, the Principal Component Analysis (PCA) provides a useful perspective on the data by establishing a new coordinate system, and it also offers a method for reducing the data’s dimensionality. Once the features have been obtained, the models are trained with WT-ARIMA-ANN. The proposed method outperforms two widely used alternatives, ARIMA and ANN.