Research On Peak Shaving Operation Of Cascade Reservoirs Under Uncertainty Of Load And Runoff Forecast

With the sustained and rapid development of economy,the power consumption has increased sharply,and the peak-to-valley difference of power load has gradually increased,which puts forward higher requirements for the stable operation of the power system.Hydropower has been widely used,and it can quickly respond to grid load commands because of its fast start and stop characteristics.Not only can it deliver high-quality energy to the power grid,it can also undertake the peak and frequency modulation tasks of the power system and provide a strong guarantee for the safe operation of the power system.The relationship between supply and demand provides a strong guarantee for the safe operation of the power system.Affected by many factors such as society,environment and humanities,the power system load and hydrological runoff have a certain degree of randomness and nonlinearity,which affects the safe and stable operation of the power system and cascade reservoirs.Obtaining high-precision load forecast results and providing high-precision runoff forecast results for cascade reservoirs are the primary key scientific problems to be solved in this study.In addition,the optimal scheduling of cascade reservoirs is affected by the uncertainty of runoff and load forecasts,which increases the risk of safe and stable operation of the power system and reservoirs.Therefore,how to effectively use the uncertainty information of runoff and load in the optimal dispatch model of cascade reservoirs is another key scientific issue in this study.This paper conducts research from the aspects of load multi-foresight period forecast,runoff multi-foresight period probability forecast,and cascade reservoir group peak shaving optimization dispatch.The research results can provide theoretical basis and technical support for cascade reservoir group peak shaving optimization dispatch under the uncertainty of load and runoff forecast.The main research work and innovation results of this paper are as follows:(1)Focusing on the key scientific issues in short-term load forecasting of the power system,a hybrid feature selection model is established and the optimal features are selected considering the relevant factors that affect the load.An improved long short-term memory network model for multi-step load forecasting is proposed.The particle swarm optimization algorithm is used to optimize the hyper parameters of the network model,which improves the prediction accuracy and training efficiency of the model.The model is applied to the research of short-term load forecasting of Hubei power grid,and compared with the current frontier forecasting model.The experimental results show that the forecasting method proposed in this paper has higher forecasting accuracy.(2)Aiming at the problem of runoff forecast and its uncertainty quantification,this paper proposes the minimal gate memory network structure based on long short-term memory network model.Introducing the pinball loss function and combining the idea of quantile regression and the density estimation method,a runoff probability prediction model is proposed.The experimental results show that the model proposed in this paper has excellent performance in the accuracy and reliability of runoff probability forecasting,and can be used for power systems.It can provide data support for the optimal dispatch of cascade reservoirs in the power system.(3)Aiming at the problem of cascaded reservoirs peak shaving optimization scheduling problem under the uncertainty of load and runoff,this paper firstly predicts the load and runoff with long foresight period probability to obtain multiple scenarios,and uses k-means clustering method to classify and obtain typical scenarios.Furthermore,with the goal of minimizing the mean square error of the reserved load of the power system,a peak shaving optimal dispatch model of cascade reservoirs is established,and the genetic evolution algorithm is used to solve the model.The case study was carried out with Qingjiang cascade hydropower stations as the research object.The optimized dispatch results show that the cascade reservoir group can effectively perform peak load regulation on the power grid and enhance the operation stability of the power system.The research results can be peak regulation dispatch under load and runoff uncertainty.The planning provides theoretical basis and technical support.