Research On Wind Farm Power Forecasting Method For Day-ahead Market

With the rapid development of China’s new energy industry and the orderly progress of electricity market construction,the proportion of wind power will continue to increase in the future and fully participate in the spot electricity market in the form of reporting volume and bidding price.The economic benefits of wind farm power generation depend more on factors such as power prediction accuracy,electricity price fluctuations,and strategy of reporting volume and bidding price.However,most of the current research on wind power prediction is how to improve the prediction accuracy,and little consideration is given to the impact of spot market price fluctuations on wind farm generation revenue.The formulation of quotation strategy is mostly based on the generation cost,electricity price prediction or game theory methods,which is conducted in an incomplete information environment,thus increasing the economic risk of wind farms.Therefore,this article focuses on the research of wind farm power prediction methods for the day-ahead market.The main research contents are as follows:(1)Day-Ahead electricity price prediction considering the output characteristics of new energyAccurate electricity price forecasting is the basis for power producers to formulate the strategy of bidding price.Because wind power has the characteristics of volatility,randomness and low marginal cost,the impact of high proportion of wind power on electricity price when participating in the day ahead market cannot be ignored.Firstly,based on the grey correlation theory,the influence of different factors on the day-ahead electricity price is analyzed,and it is found that historical electricity prices,new energy output,and load have a significant correlation with actual electricity prices.Then,a day-ahead electricity price prediction model considering the characteristics of new energy output is established using a gated recurrent neural network.The example results show that the prediction accuracy of this model is improved compared to the situation where new energy output information is not considered.(2)Wind farm power prediction for the day-ahead marketIn order to reduce the economic losses caused by the deviation of wind farm power prediction and optimize the strategy of reporting volume and bidding price,the day-ahead power generation of one’s own wind farm and other wind farms in the region are predicted.For its own wind farm,the peak time period of electricity price is first obtained according to the electricity price fluctuation,and then the day ahead wind power forecasting model considering the electricity price fluctuation is established by optimizing the loss function.The results show that the proposed model improves the forecasting accuracy of the peak time period of electricity price under the condition that the overall accuracy is basically unchanged,and ultimately improves the economic benefits of power generation of the wind farm.For other wind farms in the region,transfer learning and wind speed power curve modeling methods are used to predict their output respectively.The results of the example show that the wind speed power curve modeling method for wind farms has higher prediction accuracy,and this method can also be used to predict the power of new wind farms.(3)Strategy of reporting volume and bidding price for wind farm participation in spot marketThe strategy of reporting volume and bidding price is the foundation for power generation companies to benefit from participating in the electricity market bidding.Currently,most of the relevant research is based on generation costs,electricity price predictions,or game theory strategy.Therefore,the bidding strategy formulated based on this consideration lacks information.Therefore,based on the predicted results of the day-ahead electricity price,the power of one’s own wind farm,and the power of other wind farms in the region,a study on the bidding strategy for wind farms is conducted,and the optimal bidding strategy for wind farms is given.The results indicate that compared to wind farms bidding based on generation costs and electricity price prediction results,the profit expectation of the proposed bidding strategy is improved.