Energy Prediction And Scheduling Optimization Of Wind-solar Complementary System

With the rapid development of human society,the utilization rate of renewable energy is gradually increasing.However,both wind power and solar power have intermittency and instability problems,which pose challenges to direct power supply to the grid and meeting user demand.Therefore,this has become a major obstacle to further large-scale expansion of renewable energy.As the integration of renewable energy into the grid power increases,the energy system needs to have high flexibility to absorb more renewable energy.To meet the challenging task of enhancing China’s wind and photovoltaic consumption capacity,the integration of renewable energy and multi-energy complementary systems provides an effective solution to establish a diversified clean energy supply system.This study aims to explore the latest direction of energy development,especially the prediction of renewable energy generation power and the optimization of distributed multi-energy complementary systems.The research objective of this paper is to use stacked ensemble technology to predict short-term renewable energy generation power and optimize the distributed wind and solar complementary system at two time scales based on the prediction results.Accurate prediction of renewable energy generation power is crucial to improve the reliability and efficiency of the power system,and it can serve as a decision basis for power market transactions and distributed grid access and distribution,while the multi-energy coordinated optimization method can enhance the economic operation of the system.The main contents of this study are as follows:(1)Short-term prediction of photovoltaic power generation is crucial for the safe and economic operation of the power system.However,due to the volatility of photovoltaic power generation and its close correlation with meteorological factors,the prediction of photovoltaic power generation often has a certain degree of uncertainty.This paper proposes a novel stacked ensemble method for short-term photovoltaic power generation prediction.The method uses data and structure diversity to improve the model performance.By using the stacking method to integrate three basic models,optimizing the meta-model,and using the output of the meta-model to predict photovoltaic power generation,the proposed method is applied to the measured data of a 45 kW photovoltaic power station in Jiangsu Province under three different weather conditions.The prediction accuracy of the stacked ensemble model is the best among the three weather conditions by stacking the most basic models of the ensemble model.Moreover,the prediction accuracy of the three stacking models is higher than that of any single model,and the accuracy increases with the increase in the number of basic models in the stacking model that includes the optimal basic model.(2)Accurate short-term wind speed prediction is crucial for optimizing scheduling of wind power generation systems and maximizing profits.This paper compares the effectiveness of the power curve-based prediction method and the direct power prediction method.The indirect method first uses the stacked ensemble method to predict wind speed,and then predicts power based on the power curve,while the direct method uses the same stacked ensemble model as the wind speed prediction to directly predict wind power generation.Finally,the short-term prediction results of the total output power of the wind farm are obtained.The proposed method is applied to the measured data of a 49.5 MW wind farm in a renewable energy power generation station in Jiangsu Province.The results show that the indirect prediction method based on the power curve is slightly better than the direct prediction of wind power generation.(3)Based on predicted results,this paper proposes a dual time-scale scheduling scheme for a distributed wind-solar complementary microgrid system that includes photovoltaic(PV),wind turbines(WT),and batteries(BS).The scheduling period is divided into advance planning and real-time planning.The advance planning is based on the 24-hour advance prediction of photovoltaic and wind power,with the objective of minimizing costs to optimize intraday scheduling.Real-time planning is based on advance planning and 15-minute advance predictions of photovoltaic,wind power,and electricity load.This plan considers interactions with the main grid to solve the uncertainty of advance predictions.The operation results of the grid-connected microgrid show that the proposed rolling optimization strategy is better than the advance strategy in any scenario.Compared with the advance strategy,the proposed strategy reduces the total operating costs under sunny,cloudy,and rainy conditions and reduces the impact of prediction errors on system operation.The findings of this study provide valuable insights for China’s energy transition and efforts to increase the consumption of renewable energy.In addition,it is of great significance to the realization of China’s low-carbon and clean energy systems and dual carbon goals.