ความเป็นไปได้สำหรับการคาดการณ์แผ่นดินไหวในประเทศไทยด้วยแบบจำลองโครงข่ายประสาทเทียม.

Boonpaeng, Phanpaporn; Chaipimonplin, Tawee

Earthquake prediction with artificial neural network model (ANN) has never been studied in Thailand before. However, after searching research article from an international database, it has been found that there is a possibility of using the ANN model to predict earthquake in other countries. Therefore, this article reports on the possibility of predicting earthquakes with ANN in Thailand and collects earthquake data. The scope of this study includes three additional points; (1) The earthquake pattern in Thailand is mainly caused by the movement of active faults, also the magnitude of the earthquake and the increasing trend of earthquake rate, had a similar pattern of the occurrence in other countries when analyzing earthquake prediction with the artificial neural network model. (2) The input variables were obtained from the Gutenberg-Richter equation, which is the most input variables to be considered. The most popular output variable for earthquake prediction is the earthquake's magnitude. (3) The architectural structure design of the model mainly used Feed Forward Neural Network with Back Propagation learning. The number of hidden nodes with good performance is the ANN model with two hidden layers, also the number of hidden nodes depends on the input variables of ANN.

THAILAND; ARTIFICIAL neural networks; EARTHQUAKE prediction; EARTHQUAKE magnitude; BACK propagation; ARCHITECTURAL designs

Journal of Science & Technology MSU, 2020, Vol 39, Issue 4, p400

1686-9664

Academic Journal