Please use this identifier to cite or link to this item: http://cmuir.cmu.ac.th/jspui/handle/6653943832/69743
Title: การทดสอบแบบจำลองโครงข่ายประสาทเทียมสำหรับการคาดการณ์เหตุการณ์แผ่นดินไหวในประเทศไทยและกลุ่มรอยเลื่อนพะเยา
Other Titles: The Investigation of Artificial Neural Network Model for Earthquakes Prediction in Thailand and Group of Phayao Faults
Authors: ผู้ช่วยศาสตราจารย์ ดร.ทวี ชัยพิมลผลิน
พรรณปพร บุญแปง
Issue Date: Jun-2020
Publisher: เชียงใหม่ : บัณฑิตวิทยาลัย มหาวิทยาลัยเชียงใหม่
Abstract: This research was to study the prediction of the next earthquake in Thailand and group of Phayao faults using Artificial Neural Network model. The earthquake events were chosen to study as data for importing the model consisted of earthquakes in Thailand and in the Phayao Fault zone during the years 2015–2019 with the size of ≥ 3.5 and ≥ 3.0 magnitude, respectively. The earthquake database revealed found that Thailand occurred 51 events and group of Phayao faults occurred 32 events. These data were calculated statistically to be the input variables of the model (Mmean, dE1/2, μ, c, b, η and ΔM) and the output variable (T value). These variables were obtained from G-R relation equation. The studied variables have the units in the form of days and hours. The result suggested that the prediction of models with days as the unit of variables could predict the duration of the next earthquake in accordance with the actual data more than the model with hours as the unit of variables. Therefore, the variables in days were appropriate for the earthquake prediction of the model. In order to increase the forecasting efficiency of models with the variables in days, the architecture of models with different structures were developed, including structures with 1, 2, 3 and 4 hidden layers and the number of hidden nodes in 1, 50% and 100% of the number of input variables. Then, comparing the prediction results of each structures, this study found that the model with 3 layers and the hidden nodes of 4:4:4 were the most accurate and suitable model for earthquake prediction in Thailand. For group of Phayao faults, the most accurate and suitable model was the model with 3 hidden layers and the hidden nodes of 4:7:7. By comparing the result of the earthquake prediction in Thailand and groupof Phayao faults, it was found that the earthquake prediction in Thailand was more accurate than the earthquake prediction in group of Phayao faults, with r values of 0.899 and 0.878 respectively. The result of the next earthquake prediction suggested that Thailand will have an earthquake of ≥ 3.5 magnitude between July 1, 2020 and July 29, 2020 and group of Phayao faults will have an earthquake of ≥ 3.0 magnitude between April 26, 2020 and June 17, 2020.
URI: http://cmuir.cmu.ac.th/jspui/handle/6653943832/69743
Appears in Collections:SOC: Theses

Files in This Item:
File Description SizeFormat 
600431037 พรรณปพร บุญแปง.pdf9.16 MBAdobe PDFView/Open    Request a copy


Items in CMUIR are protected by copyright, with all rights reserved, unless otherwise indicated.