應用非優勢排序多目標遺傳演算法於水庫動態規線之研究

Investigation of Optimizing Dynamic Rule Curves of Reservoir by NSGA-II

江宜眞(Yi-Chen Chiang)

指導教授 ： 張麗秋

淡江大學/工學院/水資源及環境工程學系碩士班/碩士(2021年)

https://doi.org/10.6846/TKU.2021.00339

摘要

近年因全球暖化帶來的氣候變遷，導致極端降雨、洪水與乾旱頻繁發生。臺灣年平均降雨量為2,500公釐，是世界平均年降雨量的2.6倍，但因山勢陡峭、河川坡陡源短導致難以留下充裕的水資源。臺灣因降雨時空間分布不均，年降雨量集中在5月至9月底，為臺灣梅雨季與颱風季，乾季期間並無足夠降雨量，因此這段時期需要依賴雨季儲蓄之水資源。臺灣主要儲存水資源的方法為水庫蓄水，水庫之蓄水能力對供水能力有一定影響力，如何在極端事件降低災害損失是近年經常探討之題目。
本研究利用動態操作預測未來蓄水量並評估未來水情是否需要提前調整放流量，利用M5輸出之單旬水庫操作預測作參照組，凸顯動態操作提前限水並降低未來乾旱事件之影響程度之優點，動態操作依據未來不同水情提供最佳放流量操作策略，搜尋最佳放流量提供較安全之水庫操作。研究區域為石門水庫，本研究區分為兩種模式，分別為實際水情情境模式與極端水情情境模式，測試本模式除歷史數據外，亦測試面臨更極端事件時之模式效能，本模式設立三個目標函數作為指標，全領域搜尋最佳化操作，並將動態操作輸出之數據與歷史數據、NSGA-II操作對比分析，提供3種不同之操作建議。
本模式輸出結果顯示，本研究可預測未來乾旱程度並提供對應限水策略，提供未來水庫可操作範圍之建議，其結果顯示本模式可以降低未來乾旱嚴重程度。因歷史數據極端事件較為少數，因此本模式除實際水情情境模式外，提供了極端水情情境模式，測試模式在更嚴峻之極端乾旱時，有何種操作表現。結果顯示本模式可有效預測未來乾旱事件，提前限水以降低未來乾旱所帶來的損失與嚴重程度。

關鍵字

非優勢排序多目標遺傳演算法 ； NSGA-II ； 水庫操作 ； 動態規線 ； 乾旱 ； 數值模擬

並列摘要

Due to climate changes brought about by global warming in recent years, extreme events occurred frequently.For instance, extreme rainfall, floods, and droughts. The average annual rainfall in Taiwan is 2,500 mm, which is 2.6 times the world average annual rainfall. However, due to steep mountains and short river slopes, it is difficult to retain sufficient water resources. Due to the uneven spatial distribution of rainfall in Taiwan, the annual rainfall are concentrated on May to the end of September. It is the rainy season and typhoon season in Taiwan. There is not enough rainfall during the dry season. Therefore, this period needs to rely on water resources stored in the rainy season. Taiwan’s main method of storing water resources is reservoir storage. The storage capacity of a reservoir has a certain influence on the water supply capacity. How to reduce disaster losses in extreme events has been an important issue in recent years.
The main purpose of this research is to reduce the impact of drought events in the future, provide the best discharge operation strategy based on different future conditions, and search for the best discharge to provide safer reservoir operations.This research proves that this model is reliable by M5 operation.This study is divided into two models, namely historical data situation model and extreme data situation model. In addition to historical data, this model is also tested when faced with more extreme events. This model is set up three objective functions are used as indicators to search for optimal operations in the entire field, and compare and analyze the output data with historical and extreme data, and provide 3 different operation suggestions. The output results on this model show that this study can predict severity of drought and provide future reservoir operations. Due to there are only few extreme events in historical data, this model provides an extreme data situation model for testing the operation of the model during severer extreme droughts. The results show that this model can predict future drought events and limit outflow in advance to reduce losses and severity of disaster caused by future droughts.

並列關鍵字

Non-Dominated Sorting Genetic Algorithm ； NSGA-II ； Reservoir Operation ； Dynamic Rule Curve ； Drought ； Numerical Simulation

參考文獻

1. Anand, J., Gosain, A. K., & Khosa, R. (2018). Optimisation of multipurpose reservoir operation by coupling soil and water assessment tool (SWAT) and genetic algorithm for optimal operating policy (case study: Ganga River Basin). Sustainability, 10(5), 1660.

2. Bhesdadiya, R. H., Trivedi, I. N., Jangir, P., Jangir, N., & Kumar, A. (2016). An NSGA-III algorithm for solving multi-objective economic/environmental dispatch problem. Cogent Engineering, 3(1), 1269383.

3. Chang, F.-J., & Chen, L. (1998). Real-coded genetic algorithm for rule-based flood control reservoir management. Water Resources Management, 12(3), 185-198.

4. Chang, L. C., & Chang, F. J. (2001). Intelligent control for modelling of real‐time reservoir operation. Hydrological processes, 15(9), 1621-1634.

5. Chang, F. J., Chen, L., & Chang, L. C. (2005). Optimizing the reservoir operating rule curves by genetic algorithms. Hydrological Processes: An International Journal, 19(11), 2277-2289.