A Data-Driven Approach for Estimating Customer Contribution to System Peak Demand
share
The increasing penetration of smart meters (SMs) provides an opportunity to engage residential customers in demand response (DR) programs. The large SM datasets can be used to assess candidates' potential for DR participation. However, evaluating DR potentials in unobservable distribution systems, where a large number of customers do not have SMs, is challenging. Hence, this paper proposes a new metric to assess customers' DR potential, i.e., the coincident monthly peak contribution (CMPC). The CMPC represents an individual customer's contribution to system peak demand and can be estimated for customers without SMs. To this end, a multi-stage machine learning framework is developed: first, a clustering technique is used to build a databank containing typical daily load patterns in different seasons using the SM data of observed customers. Next, to connect unobserved customers with the discovered typical load profiles, a classification approach is applied to compute the likelihood of daily consumption patterns for different households. In the third stage, a weighted clusterwise regression (WCR) model is proposed to estimate the CMPC of unobserved customers using their monthly billing data and the outcomes of the classification module. The proposed data-driven method has been tested and verified using real utility data.
READ FULL TEXT