Physics-Aware Regression for DER Dispatch With Topological Reconfigurations of Radial Feeder

In this paper, we propose a physics-aware multi-stage regression (MSR) based algorithm to predict the power dispatches of distributed energy resources (DERs) for providing ancillary support in a smart distribution system with different topological reconfigurations. Regression collinearity is addressed with intelligent choice of the input data training set which also considerably reduces the requirements on the voltage and current measurements. Logistic regression based labeling is applied to classify the data into disjoint training sets which significantly improves the prediction accuracy. In addition, physics-aware learning is embedded with regression for predictions in different topological reconfigurations by considering switchable branches and detecting topological similarities. Simulations from 33-node 3-DER feeder and 123-node 5-DER feeder are provided to demonstrate the superior performance of the proposed algorithm in terms of accuracy, scalability and computational efficiency for voltage support application under a range of operating conditions and considering uncertainty in parameter values. The proposed approach and learnings can be extended to a range of network power flow problems and DER-based applications.