Study On Optimal Dispatching Rules Of Hydropower Reservoirs To Adopt The Load Demand Of Power Grid

In the current important development period for the promotion of energy low-carbon transformation and energy supply guarantee,hydropower with a large installed scale and mature technical system undertakes the dual mission of stable energy supply and flexible regulation of China’s power system,and the grid puts forward new and higher requirements for adaptability and efficiency of hydropower dispatching.In recent years,the power system is under great pressure to maintain power supply in summer.How to promote the efficient utilization of water resources and improve hydropower generation during the flood periods to boost the power grid tide over the peak-load during the summer period is a key issue that needs to be urgently addressed in hydropower dispatching.In addition,the negative impact of grid connection of large-scale new energy on the safe and stable operation of the power grid is becoming increasingly prominent.How to deeply explore the flexible regulation ability of cascade hydropower stations and suppress the fluctuation of new energy power and grid load is another key issue.Focusing on the above issues,this paper has conducted research work on time series forecasting in hydropower dispatching,rule extraction support method,and adaptive optimal dispatching mode of hydropower reservoir,and achieves some innovative results that can provide technical and methodological support for optimal dispatching modeling and solving for hydropower reservoirs to meet grid demands.The main research contents and achievements are as follows:(1)To explore the reasonable and effective application of time series decomposition in time series variable forecast of hydropower dispatching,A stepwise decompositionintegration-prediction framework considering boundary correction is proposed by using the stepwise decomposition sampling method,the multi-input neural network architecture,and series endpoint extension by the dynamic time warping,based on which a hybrid forecasting model of Seasonal and Trend decomposition and Long Short-Term Memory network is developed.The research results of hour-scale load forecasting of the Hubei power grid and daily runoff forecasting of the Three Gorges Reservoir indicate that the proposed model can effectively improve the forecasting performance of the single model without using future information strictly,and illustrate the reliability and practicability of the proposed hybrid framework and forecasting model,which can provide high-precision forecasting information support for optimal dispatching model and rules of hydropower reservoirs.(2)Focusing on the optimal dispatching for hydropower reservoirs under complex constraints and uncertainties,the Stochastic Fractal Search algorithm is introduced,and an improved Stochastic Fractal Search algorithm is proposed by adding the disruption update operator to solve the deterministic optimal dispatching model of hydropower reservoir.On this basis,an optimal dispatching rules extraction method of hydropower reservoirs based on the Gradient Boosting Regression Tree is proposed by adopting the framework of "optimization-fitting".The application results of daily dispatching plan compilation and dispatching rules extraction of the Three Gorges Reservoir indicate that the improved algorithm has better optimization accuracy and convergence speed than the standard algorithm,and the Gradient Boosting Regression Tree has better fitting performance in extracting the optimal dispatching rules of the hydropower reservoirs than other methods.The research results can provide method support for efficient solution of the optimal dispatching model and dispatching rules extraction.(3)In view of the power supply demand of the power grid for large hydropower stations during the summer season,the research takes the Three Gorges Reservoir as a case study,an optimal forecast dispatching model for flood resources utilization in flood season is established first based on the safety requirements of flood control and aims to promote the full and stable generation of clean hydropower during the flood season.Further considering the runoff uncertainty,a Gaussian Mixture Model is introduced for the stochastic simulation of runoff forecast errors,and then the flood control risk of flood resource utilization is evaluated.On this basis,the optimal dispatching rules of the Three Gorges Reservoir to adapt to the stable power supply demand of the power grid during the flood season are derived by using the Gradient Boosting Regression Tree.The research results show that the proposed model can promote the utilization of medium and small flood resources of the Three Gorges Reservoir during the flood season on the basis of ensuring the safety of flood control,and improve the power generation benefits and output stability of the hydropower station during the flood season.The derived dispatching rules can effectively guide the optimal power generation dispatching of the Three Gorges Reservoir during the flood season,and help the power grid to guarantee supply in summer.(4)In view of the load regulation demand of the power grid for cascade hydropower stations in the case of a large-scale new energy grid-connected,the output characteristics of wind power and photovoltaic power are first analyzed,and a short-term peak shaving dispatching model for cascade hydropower stations considering large-scale new energy integration was established.On this basis,the error probability density of wind speed forecast and solar radiation intensity forecast is estimated by using Gaussian Mixture Model,and then the stochastic simulation is conducted,and the peak-shaving optimal dispatching rules for cascade hydropower stations to meet the load regulation requirements of the power grid are deduced by using gradient boosting regression trees.On this basis,the forecast error probability density of wind speed and solar radiation is estimated by using Gaussian Mixture Model,and the stochastic simulation is conducted.Further,the optimal peak shaving dispatching rules of cascade hydropower stations that adapts to the load regulation demand of the power grid are derived by using the Gradient Boosting Regression Tree.The application research results of the Hubei power grid and Qingjiang cascade hydropower stations show that cascade hydropower stations can effectively suppress load fluctuation of the power grid and promote new energy consumption.The probabilistic decision-making output of the derived dispatching rules can provide more decision-making support information of facing time periods for the dispatcher and has practical guiding significance for the peak shaving dispatch of cascade hydropower stations in the new power system with large-scale integration of new energy.