多水庫系統最佳防洪操作之研究

Optimal Flood Control Operation for a Multireservoir System

魏志強(Chih-Chiang Wei)

指導教授 ： 徐年盛

國立臺灣大學/工學院/土木工程學研究所/博士(2006年)

https://doi.org/10.6342/NTU.2006.02606

摘要

本研究之目的為發展一多標的多水庫防洪最佳即時放水模式以及制定一最佳防洪放水規則方法流程，利用研發的即時模式和放水規則可分別得到水庫即時放水量，以提供水庫在颱洪來臨時即時放水決策之參考。
本研究為了研發水庫防洪最佳即時放水模式，提出兩不同放水策略：策略一為不考慮水庫水位指標平衡法、策略二為考慮水位指標平衡法。此兩放水策略採用混合整數線性規劃法進行模式化。策略二之模式目標函數包括下游控制點之洪峰流量最小化、水庫最大放水量最小化、水庫間水位指標差最小化以及洪水過後水庫蓄水量與目標值差最小化；限制式包括水庫連續式、水庫設施物理限制式、防洪放水原則限制式、河道演算式以及水庫水位指標平衡法限制式。本研究所決定水庫防洪最佳即時放水模式為採用兩策略下之最佳者，進而與歷史營運記錄再比較。
本研究另建立一方法流程以制定多水庫系統最佳防洪放水規則（決策樹）。所建立之方法流程包括資料收集、資料庫建置、最佳解產生、颱洪資料分類、規則萃取、最佳規則決定以及規則驗證。流程中之最佳解產生為利用定率之水庫防洪最佳全時放水模式，規則萃取則以決策樹演算法產生防洪放水規則（決策樹）。本研究所決定之多水庫系統最佳防洪放水規則為採用決策樹規則、線性迴歸規則及現行放水規則之最佳者。
本研究以淡水河流域為研究地區進行實例分析。本研究利用所收集之36場颱洪事件（1987至2004）分別進行模式分析和規則粹取之研究工作。研究結果顯示（1）水庫防洪最佳即時放水模式方面：策略二（水位指標平衡法）較策略一及歷史記錄值更能有效地降低洪峰並達到洪峰後蓄水目標。由此可知，採用水位指標平衡法之防洪即時模式可有效得到一最佳即時放水量；（2）水庫最佳防洪放水規則方面：本研究分別推求石門及翡翠兩水庫系統最佳防洪放水規則（決策樹），兩水庫之最佳決策樹規則較線性迴歸規則和水庫現行放水規則提供一更適宜之放水量在降低下游控制點最高水位及達到洪水後水庫蓄水目標值。本研究再進一步利用最佳決策樹規則改善現行放水規則。由此實際案例分析結果可知本研究成功地彰顯所提出方法之可形性。

關鍵字

防洪 ； 多水庫 ； 優選 ； 決策樹

並列摘要

The objectives of this study are: (1) development of the real-time flood control model for a multipurpose multireservoir system, and (2) building of methodology for extracting the optimal reservoir release rules for flood control. Based on the real-time model and the extracted release rules, the on-going release amounts obtained can be provided as a reference amount when operators making release policy.
To address the real-time release problem for a multireservoir flood control system, this study presents and compares two operation strategies. Strategy 1 is the real-time joint reservoir operations without using the balanced water level index (BWLI) method, while Strategy 2 involves real-time joint reservoir operations using the BWLI method. The two presented models (strategies) are formulated as mixed-integer linear programming (MILP) problems. For the model of Strategy 2, the four objectives are (1) minimizing the maximum water level at the selected control points during flood, (2) minimizing the reservoir maximum release during flood, (3) minimizing the maximal target storage error during the specific time interval at the flood ending periods, and (4) minimizing the difference between the WLIs of the two specific reservoirs in a multiple reservoir system; the constraints include reservoir continuity, physical limitations and institutional constraints, river routing, and constraints from the BWLI method. The optimal real-time model is selected from the two strategies based upon good performance and is verified by comparing with historical records.
To develop a set of optimal operating release rules for flood control, the proposed methodology is used to extract the operating rules (decision trees). The steps within the methodology include the following: (1) collection of data, (2) building of flood database, (3) generation of optimal input-output patterns by running the deterministic flood-control optimization model, (4) classification of training and testing data, (5) extraction of tree-based release rules (decision trees) for designed scenarios by using the decision-tree algorithm, (6) determination of optimal tree-based rules, and (7) verification of optimal tree-based rules through comparisons with both regression-based rules derived from multiple-linear regression model and existing release rules.
The study case is the Tanshui River Basin system in Taiwan. The collected data of 36 typhoons (1987–2004) are used for the use of the presented both real-time model and method of extracting rules. The results are described as follows.
(1) For real-time model, comparing the results obtained from the two strategies reveals that Strategy 2 performs much better than Strategy 1 in determining the reservoir real-time releases throughout the system during flood emergencies in order to minimize flooding, while maintaining all reservoirs in the system in balance if possible. Consequently, the proposed model using the BWLI method demonstrates its effectiveness in estimating on-line releases.
(2) For extracting rules, the separate tree-based rules in Shihmen and Feitsui Reservoirs are established. Results demonstrate that the solution using the tree-based rules have better performance than the regression-based rules and the existing rules in terms of reducing the peak water table at the downstream control points, and meeting the target reservoir storage at the end of flood. Moreover, this study proposes an approach for revising the existing rules according to the derived tree-based rules.
This study has successfully accomplished in formulating real-time flood-control operation model and in extracting optimal release rules in a multipurpose multireservoir system.

並列關鍵字

flood control ； multireservoir ； optmzation ； decsion tree

參考文獻

Ahmed, J. A., and Sarma, A. K., 2005. Genetic algorithm for optimal operating policy of a multipurpose reservoir. Water Resources Management, 19, 145-161.

Apte, C., and Weiss, S., 1997. Data mining with decision trees and decision rules. Future Generation Computer System, 13, 197-210.

Barbagallo, S., Consoli, S., Pappalardo, N., Greco, S., and Zimbone, S. M., 2006. Discovering reservoir operating rules by a rough set approach. Water Resources Management, 20, 19-36.

Bazartseren, B., Hildebrandt, G., and Holz, K. P., 2003. Short-term water level prediction using neural networks and neuro-fuzzy approach. Neurocomputing, 55, 439-450.

Bessler, F. T., Savic, D. A., and Walters, G. A., 2003. Water reservoir control with data mining. Journal of Water Resources Planning and Management, 129, 26-34.