Study On Short-term Water Level Prediction And Regulation Strategy Of Cascade Hydropower Stations

As a clean and renewable energy,hydropower is currently the only renewable energy that can be commercialized and applied on a large scale.Vigorously developing hydropower resources has become an important measure for countries around the world to carry out energy transformation and respond to global warming.In recent decades,my country’s hydropower energy development has entered a period of rapid development and construction.my country’s hydropower resources are mainly distributed in remote mountainous areas,alpine valleys and other areas.As more and more large and mediumsized hydropower stations are put into use,joint dispatch of cascade hydropower stations has begun to gain More and more widespread attention is that through the joint dispatch of reservoirs(hydropower stations),not only can we make full use of water resources,improve the utilization rate of hydropower resources,and increase the operating efficiency of hydropower stations,but also can implement sustainable development strategies and achieve the goals of energy saving and emission reduction.At the same time,the short-term optimal dispatching of hydropower systems continues to present a variety of challenges,one of which is the uncertainty of the short-term inflow of the hydropower station.For this reason,this article takes the Shaping II Hydropower Station in the Dadu River Basin as the actual engineering background and aims at the prediction of the water level in front of the dam.Carry out in-depth research on real-time scheduling methods.The main work and the theoretical and practical application results obtained are as follows:At present,Shaping II Hydropower Station is affected by many factors such as small water level control interval,small self-adjustable storage capacity,low unit working head,large flow during flood season and normal water period,and large unit vibration interval.The load and flood gate act frequently.In the long-term operation process,it will have a greater negative impact on the health status of the unit,flood discharge facilities and hydraulic construction.Therefore,it is necessary to build a water level prediction model to predict the water level in front of the dam,analyze the changes of the future water level in advance,and provide real-time load control strategies to reduce the load and the number of gate actions,and improve the safe operation of hydropower stations.(1)Taking the section from Zhuziba I Hydropower Station to Shaping II Hydropower Station as the research area,based on the measured water level,flow,and river topography data,HEC-RAS software is used to construct a river hydrodynamic model.Through the correlation analysis of the inbound flow of Shaping Class II and the outflow of Pillow Dam in 2019,the water flow delay time under different magnitudes is obtained.(2)Taking the water level in front of the dam,inflow and outflow of shaping II hydropower station in the whole year of 2020 as the experimental parameters and normalizing them,then BP neural network model and LSTM neural network model are selected for training,verification and testing to predict the water level in front of the dam of shaping II hydropower station in a short term.The results show that the fitting effect between the test set of LSTM model and the measured value is higher,the effect is good.(3)Based on the water level prediction model,a real-time control strategy based on the load adjustment margin is given.According to the predicted water level process,there are three situations and different decision-making logics,namely,the water level exceeds the upper limit,the water level exceeds the lower limit,and the water level does not exceed the limit.The logic can effectively reduce the gate action and load changes of the Shaping II Power Station,and ensure the safe and stable operation of the power grid.