地下水位是山坡地崩塌的決定性因素，當豪雨引起坡地地下水位上升而導致地表土層不穩定時，如能將監測所得之降雨及地下水位透過即時分析推演下一單位時間之地下水位，並建立即時預測平台，是防災議題上的重要工作。本研究採用洪耀明等(2009)所提出之地下水位即時預測模式，做為專家系統之演算分析模組，建立即時預測平台。系統中結合網路服務技術與工程資料倉儲建構線上計算平台與專家知識庫，以模型化架構開發分析模組，實作遠端線上計算，並回饋監測資料進行機器學習修正參數。雛型系統包含四部份:水文資料演算、網路服務機制、資料探勘模型及線上預測分析，各主要元件彼此獨立並相互支援，使系統更具擴充性。同時以南投縣盧山地區崩塌地降雨及地下水位資料為例，作為本研究系統之實作測試與應用。

The groundwater level represents a critical factor to evaluate hillside landslides. Whenheavy precipitation raises the groundwater level under hillslope and yields instability， it is important to generate a real time prediction platform with online analytical functions to derive the groundwater level at the next unit time step based on monitored data. The study is to present an expert system with efficient algorithms of the groundwater level fluctuation prediction model by Hong et al. (2009). A web computation platform with the MVC-based architecture is established with technology of web services and engineering data warehouse to develop online analytical modules for providing and feeding back expert knowledge online. The prototype system incorporates "hydrological data computation" ，“webservices mechanism"，“data mining model" and "online prediction analysis" to satisfy varieties of calculation requirements. The expert system is applied by rainfall data of the landslide area at Mt. Lu， Nantou County， Taiwan.