頁籤選單縮合
題 名 | 水庫系統之動態操作模式 |
---|---|
作 者 | 易任; 吳文憲; | 書刊名 | 農業工程學報 |
卷 期 | 32:4 1986.12[民75.12] |
頁 次 | 頁84-111 |
分類號 | 443.96 |
關鍵詞 | 水庫系統; 動態操作模式; |
語 文 | 中文(Chinese) |
中文摘要 | 本研究主要目的在探討水庫系統動態操作之應用,因目前一般水庫操作大部份係依據固定之操作規線,配合當時水量及下游需水量,決定該時期之放水量,此種方法雖簡捷而便於應用,但常招致水資源之若干浪費。 水庫操作為一多階段動態序率性之控制問題,欲充分利用水資源,必須能準確預測河川流量,並配合適當之水庫操作模式。本研究之內容包含兩大部分:(一)流量預測模式,(二)漸近優選模式(Progressive Optimality Alogrithm, POA)。首先以自迴歸移動平均(Autoregressive Moving Average, ARMA)模式預測流量,視流量為時間系列,以過去之流量資料,推估未來流量。繼以漸近優選模式解求動態操作問題之解答。漸近優選模式為動態規劃改良形式之一,其狀態變數不必離散化(Discretized),故可得較準確之解答,且模式未有維度問題,所需之電腦記憶容量亦小,可有效解動動態規劃之問題。 前述兩模式之運用步驟為先以流量預測模式預測一年之入流量,然後以漸近優選模式定出當年之放水量,故此研究有逐月修正誤差之效果。由本研究所確定之放水策略,當較現行之放水策略更具實用性及經濟價值,且漸近優選模式可以兼顧河水之流動時間,故於短期操作可得較準確之結果,因此可使模式適用範圍增大。 |
英文摘要 | This study is devoted to the development and application of reservoir system dynamic operation. At present most of the operating policies of a reservoir are determined by the fixed rule curves which commonly define the discharges at any time of the year and demand for water use and hydropower. Such rule curve has the advantage of concision and convenience, but it can not bring water resources into full efficience. Reservoir operation is a multistage dynamic stochastic control problem. The optimal utilization of water resource depends on the accurate forecast of inflows and the suitable reservoir operation model. This report consists of (1) Streamflow Forecast Model, (2) Porgressive Optimality Algorithm (POA). Inflow can be regarded at time series and inflow forecasts are computed by Autoregressive as time series and inflow forecasts are coputed by Autoregressive Moving Average (ARMA) model. The POA is a modification of Dynamic Programming (DP). This method is computationally efficient and has minimal storage requirements. Contrary to DP, the state variables do not have to be discretized with this method. At every beginning of periods (e. g., month), the forecast of inflow is made by ARMA model, and release policies are determined by POA. As inflow forecast is deviated from actual realizations, new forecast is computed and a revised future release policy is developed on the basis of the observed state of the system and update forecast. From this research, we obtain the conclusion that this model is superior to current operating policies. And POA can take into account the time of water travel between upstream and downstream reservoir. Therefore this model can be suitable for most cases. |
本系統中英文摘要資訊取自各篇刊載內容。