A Research On Time-of-use Pricing Mechanism Of Electricity Retail Market Based On Demand Response

With the rapid development of China’s power system and further advancement of market-oriented reform of the power sales side,the current pricing method of the retail electricity market has the problem that it cannot reflect the market supply and demand situation.In the power market where the power sales side is liberalized,the pricing mechanism of electricity retailers plays an important role in effectively regulating the balance of supply and demand.Therefore,researching the time-of-use pricing mechanism based on demand response has important theoretical significance for the advancement of China’s power system reform and the rational allocation of power resources.A demand response model based on an AT-Bi LSTM-LSTM neural network was established.On the basis of analyzing and proving that there was a strong correlation between electricity price and electricity demand,this paper established a demand response model based on an AT-Bi LSTM-LSTM neural network.The model first extracts the features based on the factors affecting the power demand,and then uses the Bi LSTM neural network layer for learning bidirectional time-series features.Then we calculate different weights of the LSTM hidden layer state by attention mechanism to focus on the input features selectively.Combining the attention weight and the LSTM neural network to perform demand prediction.In this paper,the electricity data set is used for experiments.Compared with others typical forecasting models,proposed AT-Bi LSTM-LSTM neural network method produces more accurate and effective predictions.At the same time,the uncertainty of power demand was considered in this paper,the demand probability density was used to describe the demand uncertainty,and a demand response probability model based on an ATBi LSTM-LSTM-QR neural network was established.The experimental results showed that the demand response probability model proposed in this paper has a higher prediction interval coverage probability and a smaller prediction interval width.Based on the demand response model and Stackelberg game,the time-of-use pricing mechanism was studied in this paper.With retailers and power users as the main players in the game,we use the demand response model to replace the user utility function in the traditional game model,and establish the Stackelberg time-sharing pricing game model.At the same time,we give an optimization framework for solving the time-of-use pricing mechanism.The experiments have proved the time-of-use pricing mechanism proposed in this paper can effectively reduce the daily load deviation,reduce the cost risks,improve the user satisfaction and increase the retailers’ expected profits.At the same time,this paper established a Stackelberg time-of-use pricing game model with incomplete information in view of the uncertainty of demand response,and the demand response probability model was used as the user’s utility function.The experiments have proved that time-of-use pricing mechanism developed using this model can further reduce the retailer’s cost risk.In addition,this paper also researched the influence of retailers’ risk preferences on the results of the time-of-use pricing decision,and provided some relevant conclusions.