Customer Baseline Load Bias Estimation Method of Incentive-Based Demand Response Based on CONTROL Group Matching

Abstract

Accurate customer baseline load (CBL) estimation is of great significance for demand response (DR) performance evaluation and financial settlement of DR participation rewards. However, due to customers' random electricity consumption behaviors, the CBL estimation errors are unavoidable. Bias is usually used to quantify the CBL estimation error, which provides the basis for DR program operators to select the most appropriate CBL model and optimize the DR program. Unfortunately, it is impossible to meter the actual bias in practice because the actual CBL would never exist once the DR program is implemented. In this paper, a CONTROL group matching based approach is proposed to estimate the CBL bias. All customers are divided into DR and CONTROL group, including DR participants and non-DR customers, respectively. The basic idea is that use the bias information of those CONTROL customers who don't participate DR program but show similar bias distribution with the DR group in the historical days prior to DR event day to estimate the bias of DR group on the DR event day. A case study using a dataset of more than 4000 residential customers shows that the proposed approach has better overall performance than other benchmark methods.