基于多算法耦合的水电站自主智能优化调度应用  

Application research on autonomous intelligent optimal dispatching of hydropower station based on machine learning

作　　者：王向伟 WANG Xiangwei(Datang Xiangcheng(Derong)Hydropower Development Co.,Ltd.,Chengdu 610000,China)

机构地区：[1]大唐乡城(得荣)水电开发有限公司,四川成都610000

出　　处：《技术与市场》2022年第12期7-11,共5页Technology and Market

基　　金：国家重点研发计划“长江上游梯级水库群多目标联合调度技术”(2016YFC0402208)。

摘　　要：中、大型水电站的水位预测以及优化调度是水库防洪和发电能力的重要问题,目前对水库水位预测以及水库调度主要采用算法预测水位与人工设置调度模式相结合的方式,使得水库智能调度模式在实际的调度过程中实用性不强。为满足水电站能自主智能优化调度,脱离人工设置,利用大渡河流域某电站水库水位数据建立了基于BP(back propagation)神经网络水库水位预测模型。采用C4.5算法模拟调度人员对水电站调度模式选取规则,再利用POA(逐步优化)算法模型通过耦合BP神经网络水位预测模型,对电站水库预测的水位以及C4.5机器算法选取调度模式的规则,以达到对水电站自主智能优化调度。解决了POA算法须人工设值调度期末控制水位与调度过程优化模式的缺陷,使得多算法耦合模型可以脱离人工设值自行执行符合电站运行策略的优化操作,同时验证了多算法耦合的可行性和有效性。The water level prediction and optimal operation of medium and large hydropower stations are important problems of reservoir flood control and power generation capacity.At present,the combination of algorithm prediction of water level and manual setting of operation mode is mainly used for reservoir water level prediction and reservoir operation,which makes the intelligent operation mode of reservoir not practical in the actual operation process.In order to meet the independent and intelligent optimal dispatching of hydropower station,it is separated from manual setting.In this paper,the reservoir water level prediction model based on BP(back propagation)neural network is established by using the reservoir water level data of a power station in Dadu River Basin.Use C45.The algorithm simulates the selection rules of dispatching mode of hydropower station by dispatching personnel.Then,the POA(step-by-step optimization)algorithm model is used to predict the water level of the power station reservoir and C45.The machine algorithm selects the rules of dispatching mode to achieve the autonomous and intelligent optimal dispatching of hydropower station.It solves the defect that POA algorithm needs to be set manually to control the water level at the end of the dispatching period and the optimization mode of the dispatching process,so that the multi algorithm coupling model can separate from the manual setting and perform the optimization operation in line with the operation strategy of the power station.At the same time,it verifies the feasibility and effectiveness of multi algorithm coupling.

关 键 词：BP神经网络 C4.5算法 多算法耦合 优化调度 

分 类 号：TV736[水利工程—水利水电工程]