Earth Quack Prediction of Southern California Using Unsupervised Clustering Based Ensemble Deep Learning Model

Predicting when and where an earthquake will occur is a persistent topic in seismology that has attracted the interest of researchers and the general public. There is currently no physical or statistical model that can reliably anticipate the occurrence of major earthquakes, despite continuous research into the fundamental factors that cause them. Random forests and long short-term memory (LSTM) networks are two examples of machine learning techniques that show promise for enhancing earthquake prediction accuracy because of their ability to quickly discover patterns in vast databases. An unsupervised ensemble learning model trained on a labeled dataset has been built for earthquake detection in this article. An unlabeled dataset can be transformed using an ensemble machine learning strategy, which is the result of combining clustering methods with weighted voting. The Adaptive Tunicate Swarm Optimization Algorithm (ATSOA) is used to select the LSTM's weight. The dataset gathered by the Southern California Seismic Network is used for the method's training and testing. The results show that the suggested technique has a high potential for accurate seismic signal detection and phase identification. The provided approach outperforms other suggested deep learning algorithms on the Southern California Earthquake Data Center's recorded dataset by around 4% in earthquake detection and by a small margin in seismic phase identification. Furthermore, the experimental result shows that the model is reliable in recognizing low-magnitude seismic phases. Finally, we conclude that the suggested learning provides a viable option for enhancing earthquake occurrence prediction, and that minor earthquakes include information useful to forecasting future maj or earthquakes.