Stochastic Dynamic Programming And Value Evaluation In Reservoir Optimal Operation

As a modem water resource management strategy, Reservoir optimal operation can improve level and utilization efficiency of water resources and reduce nonrenewable energy consumption, which promotes the sustainable development of the social economy. Establishing an accurate optimal operation model and making reasonable schedule rules are basic work of hydropower station. At the same time, evaluating value of the model mentioned above objectively and having a clear promoting space of optimization dispatching benefit are helpful to enhance the level of reservoir dispatching management.Traditional value evaluation method of water conservation and generation increasing usual y assess and evaluate the generation benefit for a previous year at year's end. The assessment result can't express the value of a dispatching model and the expected promoting space of optimization dispatching benefit. Therefore, this dissertation first establishes stochastic dynamical programming model to get dispatching rules of a hydropower station, and then introduces the concepts of expected value of perfect information(EVPI) and value of stochastic solution(VSS) to estimate the optimizing degree of dispatching model and promoting space of expected value in reservoir operation. On the above research idea, the idiographic work is as follows:(1) Inflow description. This dissertation makes correlation analysis on inflow processes, describe the inflow processes in different ways, introduces computing methods of representative inflow and there probabilities, analyzes the relationship between different description ways and discusses the form of transition probability matrix of deterministic and intervals independent description ways. A general inflow description way is proposed based on above.(2) Optimal operation model in reservoir operation. This dissertation describes inflow processes by interval independent and dependent description ways, establishes stochastic dynamical programming model with taking these inflow processes as input and the goal of maximization generation and gets dispatching rules. The simulations reveal that the application of SDP models gain better generation benefit than regular schedule. Compared intervals dependent SDP model with intervals dependent SDP model, the later acquires more generation benefit. Thus, the stochastic dynamical programming model and its dispatching rules are effective.(3) Value evaluation of stochastic dynamical programming model in reservoir operation. This dissertation introduces the concepts of expected value of perfect information(EVPI) and value of stochastic solution(VSS) to estimate the value of inflow and models in reservoir operation for the defect of conventional methods. The results evaluated reflect the value of inflow and stochastic dynamical programming model wel, define the optimizing degree of stochastic dynamical programming model and the room for improvement of reservoir optimal operation and provide evidence for promoting operators to have a better management of hydropower.