運用馬可夫鍊隨機場模擬與合成地質模型評估地質模型不確定性

NCU Institutional Repository > 地球科學學院 > 應用地質研究所 > 博碩士論文 > Item 987654321/86928

請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/86928

題名:	運用馬可夫鍊隨機場模擬與合成地質模型評估地質模型不確定性
作者:	謝澤銘;MENG, CHE CHAK
貢獻者:	應用地質研究所
關鍵詞:	合成地質模型;地質模型不確定性;不確定性量化;地質知識;亂度;最大概似估計;Synthetic geological model;Geological model uncertainty;Uncertainty quantification;Geological knowledge;Entropy;Maximum Likelihood Estimation
日期:	2021-10-13
上傳時間:	2021-12-07 13:26:57 (UTC+8)
出版者:	國立中央大學
摘要:	建構符合現地狀態之地質模型(Geological Model)將有助於土木、大地、地下構造物、壩體等工程設計依據，並且可提供地質相關災害之分析與風險評估。藉由各式現地取樣資料建構之地質模型，由於經費限制或其他因素導致資料量的多寡直接影響地質模型之擬真程度。或因工程師地質專業及使用套裝軟體內插之侷限性，各式地質模型將具備其各種程度之不確定性(Uncertainty)。本研究先建立合成地質模型作為資料基礎，然後取樣模型資料進行馬可夫鍊隨機場(Markov chain random field)計算，進而探討評估地質模型不確定性之方法。以場址為嘉義縣民雄鄉民雄、頭橋工業區附近區域為模型研究案例，使用地質鑽探岩心資料為基礎，由實際現地資料建構合成(Synthetic)地質模型，搭配前人研究在此區域的地質研究和結合地質知識，建立出符合地質學邏輯的合成地質模型，然後取樣模型資料，進行馬可夫鍊隨機場的敏感性分析，從中考慮如何設計馬可夫鍊隨機場的模型參數，進而量化地質模型之亂度(Entropy)以表示地質模型不確定性，利用最大概似估計(Maximum Likelihood Estimation)和地層邊界殘差分析呈現模擬模型的準確性，以評估模擬模型之最佳參數，找出地層傾斜角度和材料側向延伸性。最後從結果討論，亂度量化的不確定性反映的是資料計算的先天變異性，而從最大概似評估在地質條件參數組合都能達到最佳值的結果來看，最大概似評估為最能評估地質學原理對地質模型不確定性之貢獻的方法。;Simplifying the geological model that conforms to the current state of the ground will help the engineering design basis of civil engineering, geotechnical engineering, underground structures, and dams, and provide analysis and risk assessment of geological disasters. However, there are some difficulties with constructing geological models. For example, there are few data collected from samples or the limitation of engineers and software. These reasons all affect the uncertainty of model. In this study, a synthetic geological model is established as the data basis, and then the model data is sampled for Markov chain random field calculation, and then methods for evaluating the uncertainty of the geological model are discussed. Taking the site is Minxiong Township, Chiayi County, and the vicinity of Touqiao Industrial Park as a model study case. Based on geological drilling core data, a synthetic geological model is constructed from actual on-site data, combined with previous research here. Regional geological research and geological knowledge are combined to establish a synthetic geological model that conforms to geological logic. After sampling the model data to conduct sensitivity analysis of the Markov chain random field, consider how to design the model parameters of the Markov chain random field. Quantify the entropy of the geological model, evaluate the Maximum Likelihood Estimation of the geological model to show the uncertainty of the geological model, use the maximum likelihood estimation and the residual analysis of the stratigraphy boundary shows the accuracy of the simulation model to evaluate the effect of the best parameters. Find out the inclination angle of the formation and the lateral continuity of the material. Finally, discuss how geological principles can reduce the uncertainty of the geological model.
顯示於類別:	[應用地質研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	193	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....