利用模擬退火演算法推估地下污染物來源;Identification of the groundwater contaminant source using simulated annealing algorithm

NCU Institutional Repository > 地球科學學院 > 應用地質研究所 > 博碩士論文 > Item 987654321/85124

請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/85124

題名:	利用模擬退火演算法推估地下污染物來源;Identification of the groundwater contaminant source using simulated annealing algorithm
作者:	鍾佳良;Jhong, Jia-Liang
貢獻者:	應用地質研究所
關鍵詞:	模擬退火演算法;地下水污染溯源;系集;simulated annealing algorithm;groundwater contaminant source identification;ensemble
日期:	2021-01-28
上傳時間:	2021-03-18 17:43:10 (UTC+8)
出版者:	國立中央大學
摘要:	地下水被視為穩定且可靠的用水來源，卻也因其不可見及水循環更新速度慢這兩特性，一旦受到污染，管理單位難以在第一時間發現，也導致污染源會隨時間推移而擴大其分布範圍，直接或間接增加民眾用水品質風險。在面對受污染的水體，需要瞭解污染傳輸歷程。要解污染物之傳輸歷程，須瞭解該地區的水文地質條件及地下水污染傳輸等機制，方得以數值模式進行相關推估，基於各項參數取得條件之不易、監測井數量有限，推估的工作變得相當困難。為解決現地觀測資料的不足所帶來之影響，本研究開發出以python介接數值模式與啟發式演算法(heuristic algorithm)推估污染傳輸之方法，其中水文及污染傳輸數值模式分別使用MODFLOW-2005與MT3D-MS，用於生成測試案例與當作最佳化問題的數值模式。啟發式演算法則採用退火演算法(simulated annealing algorithm, SA)，用於求解最佳化問題之演算法。為減少推估誤差與提升系統之穩健性，本研究將數值模式中所使用的參數設定成一特定範圍並離散化，降低因固定參數所產生之誤差；對於退火演算法及動力系統所帶來的不確定性，利用系集(ensemble)與空間分析之概念，降低單一次模擬所帶來的誤差，提高推估系統穩健性。為驗證推估方法，研究中使用均質地質條件與單一且持續性的污染點源釋放，合成兩假想之污染傳輸案例作為推估方法的驗證，經測試案例驗證後，推估方法除能夠有效推估污染源之點位、傳輸方向，本研究於方法建立後，探討並量化不同數量的觀測點對於此方法的適用性與不確定性，結果顯示在監測井數量為6以上時，點源推估之誤差皆小於1公尺，傳輸方向誤差為5度以內。;Groundwater is one of the stable and reliable water resources. Due to the natural complexity of the aquifer systems, early detection of the contaminant is challenging. The spread of the contaminant in a groundwater system might directly or indirectly influence residents near the contamination sources. Typical approaches to predicting the migration of the plumes are the numerical models, which allow the reproductions of the transport processes and quantify the concentration distribution. Due to the limited parameters for flow and transport models, the implementations of most numerical models become complicated. We introduced a heuristic algorithm to estimate flow and transport parameters for a contaminant site in the study. The model tests consider two numerical examples, where the flow directions and numbers of observations are different in the designed scenarios. Specifically, the MODFLOW-2005 and MT3D-MS models were used for generating the observations at the synthetic sites. The MODFLOW-2005 and MT3D-MS models associated with the simulated annealing (SA) algorithm were developed and systematically tested to quantify the accuracy of the estimated flow and transport parameters. In the study, the numerical model parameters were fixed to a specific range and discretized to reduce the estimation errors and improve the system′s robustness. Additionally, the ensemble approach and spatial analysis were used to reduce the errors caused by a single simulation. Results showed that the developed model could effectively identify the locations and directions of the contaminant source. The model enables evaluating suitable observation wells for the sites and quantifies associated uncertainty induced by the limited observations. The numerical examples showed that the number of monitoring wells needs to be higher than 6. The source point estimation error can be less than 1 m, and the error of the transport direction can be within 5 degrees.
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	135	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.....