多標的遺傳演算法探討南化水庫最佳限水策略

Exploring Optimal Hedging Rules of The Nanhua Reservoir Using Multi-Objective Genetic Algorithm

黃景裕(Jing-Yu Huang)

指導教授 ： 張麗秋

共同指導教授 ： 蕭政宗(Jenq-Tzong Shiau)

淡江大學/工學院/水資源及環境工程學系碩士班/碩士(2008年)

https://doi.org/10.6846/TKU.2008.00265

摘要

本文研究目的為利用多標的遺傳演算法探討南化水庫乾旱時期最佳限水策略，限水策略以標準操作策略(SOP)為基礎並加入限水參數，限水策略依參數個數分為一點法、二點法、及三參數法，另依參數是否隨時間變化分為定值及時變限水策略，所考慮的參數時間變化頻率有半年變化、季變化及月變化。本文選用相互衝突的總缺水率與單旬最大缺水率作為衡量供水水庫營運效率的指標，並以非優勢排列遺傳演算法(NSGA-II)求解以此二缺水指標為標的函數的多標的Pareto最佳解。經應用於南化水庫分析後顯示增加限水參數個數及參數時間變化頻率可有效改善水庫限水效果，即 Pareto鋒線往減少總缺水率及單旬最大缺水率的方向移動，且其限水效果可相互疊加，因此在所分析的十二種限水策略中以三參數法月變化限水策略為最優。

關鍵字

多標的遺傳演算法 ； Pareto最佳解 ； 限水策略 ； 缺水率 ； 可利用水量

並列摘要

This study aims to exploring optimal hedging rules using multi-objective genetic algorithm for the Nanhua Reservoir during droughts. Hedging parameters are added in the SOP-based rules to construct water-rationing measures. One-, two-, and three-parameter hedging rules associated with constant and time-varying hedging parameters are employed to investigate effects on water-shortage characteristics. Time-varying frequencies considered in this study include semi-annually, quarterly, and monthly varying. Two conflicting shortage indices, total shortage ratio and maximum 10-day shortage ratio, are used to evaluate operation performance of a water-supply reservoir. The Pareto optimal solutions of this multi-objective optimization are searched by the non-dominated shorting genetic algorithm II (NSGA-II). The proposed methodology is applied to the Nanhua Reservoir that is located in southern Taiwan. The results show that increasing time-varying frequency of hedging parameters can effectively reduce water-shortage characteristic, which are further improved by increasing numbers of hedging parameters. Thus, the three-parameter monthly varying hedging rule performs best among twelve hedging rules evaluated in this study.

並列關鍵字

multi-objective genetic algorithm ； Pareto optimal solutions ； hedging rule ； shortage ratio ； water availability

參考文獻

1.Bayazit, M, and Unal, E., “Effects of hedging on reservoir performance,” Water Resources Research, Vol. 26, No. 4, pp. 713-719 (1990).

2.Bekele, E. G., and Nicklow, J. W., “Multiobjective management of ecosystem services by integrative watershed modeling and evolutionary algorithms,” Water Resources Research, 41, W10406, doi: 10.1029/ 2005WR004090 (2005).

3.Burn, D. H., and Yulianti, J. S., “Waste-load allocation using genetic algorithms,” Journal of Water Resources Planning and Management, ASCE, Vol. 127, No. 2, pp. 121-129 (2001).

5.Deb, K., Multi-objective Optimization Using Evolutionary Algorithms, John Wiley & Sons, Chichester, pp. 389-400 (2001).

7.Draper, A. J., and Lund, J. R., “Optimal hedging and carryover storage value,” Journal of Water Resources Planning and Management, ASCE, Vol. 130, No. 1, pp. 83-87 (2004).