Research On Real-time Dispatching Strategy Of Shaping Ⅱ Power Station

Shaping Ⅱ Hydropower Station located in the middle of the Dadu River cascade has a small water level control interval,a low unit working head,a large flow during flood and normal water periods,and a large unit vibration interval.Under the influence of various factors,the load fine-tuning is frequent and the flood gate operates frequently.In the longterm operation process,it will have a greater negative impact on the health of the unit,the flood discharge facilities and the hydraulic construction,and the cumulative and frequent irregular actions are very likely to cause damage to the equipment and facilities.As climate change,human activities and other factors,intermittent energy such as large-scale wind and solar are fed into the grid,the uncertainty of grid load and the uncertainty of interval water are intensified,and the load plan caused by multiple uncertain factors is frequently adjusted,The problems of load plan adjustment deviation caused by gate opening and water abandonment and frequent gate opening and closing have become common problems and difficulties faced by the real-time dispatching and operation of cascade hydropower stations in China.This paper takes the Shaping Ⅱ Hydropower Station as the research object,focuses on the key issues faced in the real-time dispatching process,and uses theories and methods such as data analysis,river dynamics,deep learning,and hydropower energy as the guidance to explore the rules of real-time dispatching operation and perform super Short-term reservoir water balance simulation,establishment of an ultra-short-term water level prediction method in front of the dam,and formulation of a real-time dispatch strategy for Shaping Ⅱ Hydropower Station,which solved the problem of real-time dispatch optimization model in practical engineering applications.The main research contents of this paper are as follows:(1)Introduced the basic data of Shaping Ⅱ Hydropower Station in detail,and sorted out the specific process of current real-time dispatching.Summarize the current problems faced by the real-time dispatching of the Shaping Ⅱ Hydropower Station,and analyze the reasons for the difficulty of real-time dispatching of the Shaping Ⅱ Hydropower Station.(2)Using theoretical derivation and statistical analysis methods to carry out research on the time distribution and change law of real-time dispatching data,response mode,and action process,and establish a method and system for accurate description of real-time dispatching operation law analysis;revealing the historical real-time dispatching of Shaping Ⅱ Hydropower Station Operation law;a one-dimensional hydrodynamic model of river channel-reservoir area was established,and the law of river runoff evolution was analyzed.And on this basis,the accurate simulation of the ultra-short-term reservoir water balance model is realized,which provides theoretical foundation and data support for the ultra-shortterm water level prediction in front of the dam and the modeling of the real-time dispatch model of the Shaping Ⅱ Hydropower Station.(3)In view of the complexity,nonlinearity,parameter uncertainty,and disturbance randomness of the water level in front of the dam,an ultra-short-term water level prediction model with a variety of deep learning algorithms as the core is established.Common accuracy evaluation indicators are used to compare and analyze the error accuracy of all models,and an error analysis method for real-time prediction of the ultra-short-term water level in front of the dam is proposed,and the applicability of different accuracy evaluations to the ultrashort-term water level prediction model in front of the dam is demonstrated.Choose a model with high accuracy and good robustness from each model.The research results show that the average absolute percentage error and the coefficient of determination are not suitable for the evaluation index of the ultra-short-term water level prediction model before the dam.After error comparison analysis,the BP neural network model has high accuracy and good robustness.(4)Taking the real-time dispatching requirements of different working conditions and different optimization goals in the real-time dispatching process of Shaping Ⅱ Hydropower Station as the starting point,a real-time dispatching strategy for Shaping Ⅱ Hydropower Station was formulated.The research work fully considered the multiple uncertain factors faced in the actual dispatching process,established a real-time dispatching model for Shaping Ⅱ Hydropower Station with optional optimization targets and constraint conditions,and formulated detailed procedures including load applications and gate opening adjustments.Real-time dispatching strategy for Shaping Ⅱ Hydropower Station.Construct the water consumption rate model and the water level model to optimize the simulation of the selected16 working conditions,and use the 2019 data for long-term optimization simulation.The optimization simulation of working conditions shows that when the inbound flow is stable,the scheduling process is relatively simple,and the optimization space is not large.When the inbound flow is turbulent,if the gate is not opened,the optimization effect of the water level model is better,the water level is stable,and the water consumption rate is low.When the gate is opened,the water level model is no longer applicable,and the water consumption rate model can be used,but the constraint conditions under specific working conditions must be handled correctly.Long-term optimization simulation shows that the optimization model has a significant effect on the optimization of Shaping’s real-time dispatch process,the water level process is more stable,and the economic benefits are significant.In the end,a trial operation was carried out at the Shaping Ⅱ Hydropower Station,and the set optimization goals were completed,with good operating results.