應用智慧型演算法於地表地下最佳聯合營運模式之建立

Abstract

近年來由於全球氣候變遷造成旱澇等極端氣候發生的頻率升高，且水資源需求 又日益增加，若僅依靠地表水源將增加穩定供水的風險，而相較於地表水源，地下 水的供水穩定，若能將地下水與地表水進行聯合營運，則將可大為提高水源供應的 穩定度。雖然，台灣部分地區已過度抽用地下水，然而地表地下聯合營運的目的並 非在增加地下水之使用量，它的真正精神是利用地下水層龐大的儲存空間，進行地 表水的調蓄操作，因此在適當的營運下，不但可推持地下水的進出平衡，甚至可進 一步達到補注涵養地下水的功能。 本研究目的在於建立一個可正確描述非穩態線性與非線性地下水系統之地表地 下聯合營運管理模式，傳統上常見地下水之實體結構物有兩大類，分別為「水井群」 與「人工湖」，實際應用上可能是獨自運用，或是混合運用於地表地下營運中。因 此本研究首先將應用類神經網路與遺傳演算法，分別針對不同型式之實體結構物建 立聯合營運模式，最後再綜合前兩年度之結果，結合模糊控制理論建置一個混合型 式之聯合營運模式，藉此本研究所發展的模式將可建議出包含水井與人工湖設施的 最佳地表地下聯合營運規劃策略，在永續經營的前提下，有效提升實際水資源營運 效率，降低缺水風險。

Due to the fast increasing of water demand and the raising hydrological variation caused by the global climate change, using only the surface water will increase the risk of water shortage in Taiwan. Comparing to the surface water, the groundwater resource is much more stable than the surface water. Therefore, the conjunctive-use of surface and sub-surface water can greatly reduce the water shortage risk. Although there are places that the groundwater was over pumped in Taiwan, the situation does not prohibit the implementation of conjunctive-use since the merit of the conjunctive-use is not to use more groundwater resource but to utilize the aquifer as a new storage system of the surface water to increase the capacity and efficiency of surface water allocation in time. The conjunctive operation is not only to preserve the balance of groundwater recharge and pumping. Beside, depending on the operation objective, the conjunctive operation can even recharge the aquifer to conserve the groundwater resource. This study is to develop a optimal planning model for the conjunctive-use of surface and subsurface water. The proposed model is capable of considering both linear surface water systems and nonlinear groundwater systems simultaneously. Conventionally, wells system and artificial lake are two hydraulic structures generally used in the groundwater management, and both structures can be applied simultaneously or independently in the conjunctive-use system. Therefore, the research consists of three steps (years). The first and second years are to develop the conjunctive-use models that apply either the well system or the artificial lake for groundwater supply. The final year is to integrate the two years results and develop a general conjunctive-use model that considers both the well system and artificial lake. For the beginning two years, the Genetic Algorithm (GA), Artificial Neural Network (ANN) and Linear Programming (LP) are adopted to develop the planning model. This study further applies the fuzzy control theorem to allocate the water demand between the surface and sub-surface system in the third year. The proposed model can analyze different planning strategies for the conjunctive-use of surface and sub-surface water and obtain a strategy which can fulfill the requirement of sustainable management, increase the system operating efficiency and reduce the water shortage risk.