透過您的圖書館登入
IP:3.144.35.148
  • 學位論文

應用專家系統於穩健型地下水參數檢定模式之發展

Applying Expert System on the Development of a Robust Model for Groundwater Parameter Identification

指導教授 : 張良正

摘要


傳統以數學優化法為基礎的地下水參數優選模式如 UCODE 等, 在參數維度高時,常易產生不易收斂的問題,有鑑於此,本研究乃結 合專家系統與MODFLOW 2000 地下水數值模式,以專家之檢定規則 為基礎,發展強健型參數檢定方法,克服傳統自動化檢定方法於高參 數維度時之收斂問題。由於專家系統是以檢定規則訂定檢定策略,故 使用者可有彈性地增減或修正檢定規則,進而擴充及完善其檢定功 能。本研究目前以暫態淨補注量作為檢定對象,並將本模式應用於濁 水溪沖積扇地下水模擬,且與常見之自動參數檢定模式(UCODE)進行 比較。 經較簡單的設計案例與濁水溪沖積扇實際案例之驗證後,本研究 之參數檢定系統不論是應用於設計案例或是複雜的現地濁水溪沖積扇 地下水暫態模擬,均可檢定出合理的淨補注量,可證實本系統之正確 性及實用性。此外,在高維度之現地案例中,相較於UCODE 嚴重受 到初始猜值之影響,本模式充分展示其穩健求解之能力,因此可為地 下水參數檢定提供一穩健之解決方案。

並列摘要


Conventional groundwater parameter identification modeling based on optimization, such as UCODE, has difficulty converging to a global optimum in a high dimension situation. To avoid this convergence problem, this study integrates a rule-based expert system and a groundwater simulation model, MODFLOW 2000, to develop a robust methodology for groundwater parameter identification. Because the expert system requires calibration rules to identify parameters, users can flexibly add new rules or modify existing rules with this proposed methodology. Therefore, the proposed methodology can adapt for new parameter identification problems easily. We apply this proposed methodology to a real case study of Choshuihsi Alluvial Fan which is located at the central Taiwan. To test the robustness for high dimension problems, the proposed methodology is applied to calibrate the net recharge rates in a transient simulation in the study area. The result is compared with the calibration results obtained from UCODE. The results show that the initial guess dramatically effects the convergency of the optimization using UCODE, but our proposed methodology is very robust for achieving the convergence requirements of output error criteria for high dimensional problems. These results presented the robustness and the applicability of the proposed methodology for high dimensional groundwater parameter identification problems.

參考文獻


31. 王雲直,應用專家系統於地下水模式參數檢定之研究─以濁水溪
30. 陳韋圻,應用專家系統於地下水模式自動化參數檢定之研究,國
18. 巨廷工程顧問股份有限公司、國立交通大學,多元化水資源經營
5. Sang Min Kim, Brian L. Benham, Kevin M. Brannan, Rebecca W.
1. K.W. Chau, “Intelligent manipulation of calibration parameters in

被引用紀錄


楊深惠(2012)。應用衛星影像辨識與河道水理演算於濁水溪沖積扇地下水數值模擬〔碩士論文,國立交通大學〕。華藝線上圖書館。https://doi.org/10.6842/NCTU.2012.00389

延伸閱讀