Data-driven Demand Response Potential Prediction For Business Users

In order to achieve the "double carbon" goal as soon as possible,in addition to seeking solutions from the supply side,the demand-side resources also contain huge potential.At present,the regulation potential of the demand side has not been fully utilized and has not been fully mobilized by the power grid for normalization regulation.Building a new power system,promoting the consumption of renewable energy and improving the level of social energy efficiency all need to mobilize users to participate in the power grid regulation.In this context,the adjustable potential of commercial load needs to be deeply explored,and the analysis of its adjustable potential is of great significance to alleviate the contradiction between supply and demand of power grid,and to promote the value mining of demand-side resources and the implementation of regulation mode.Commercial users have objective adjustment potential,but the existing research has not carried out adjustment potential analysis for their electricity consumption behavior and seasonal preferences,which makes it difficult for commercial users to participate in demand response.In view of the above problems,the main research results of this paper include the following points:(1)Analyze the load characteristics of the commercial complex,analyze the regulation and control of the main electrical equipment,and propose the corresponding peak shifting operation potential indicators.(2)A method for evaluating the potential of commercial users is proposed.Based on the improved K-means clustering algorithm,the prediction potential of commercial users is combined with historical data to carry out a classified evaluation of the adjustment potential of the seasonal dimension,and the off-peak operation potential of users’ seasonal preferences is obtained.(3)Propose an analysis and calculation method for the adjustable potential of commercial users.Considering the characteristics of the adjustable potential of commercial users,use the semi-supervised Self-Training model to expand the data set,and train the expanded user data based on Bi-LSTM neural network to obtain the adjustable potential of commercial users.The research content of this paper is conducive to the friendly interaction between the power grid and demand-side resources,can divide the potential of user load regulation according to the time and people,ensure the accuracy and rationality of the implementation of demand response regulation,help match the commercial load response characteristics and demand response scenarios,and help the construction of new power systems.