Applying Networked Estimation and Control Algorithms to Address Communication Bandwidth Limitations and Latencies in Demand Response

Abstract

Demand response can provide services to the power network, however, coordination of spatially distributed demand response resources generally requires coping with imperfect communication networks. This work investigates methods to manage communication constraints (e.g., Delays and bandwidth limitations), faced by demand response aggregators who manipulate the on/off modes of residential thermostatically controlled loads (TCLs). We present two model predictive control (MPC) algorithms that exploit a priori knowledge of delay statistics. We also present three Kalman filter-based state estimation methods that handle measurements with heterogeneous delays that are known a posteriori. We simulate the closed loop system to quantify the error while the system tracks simplified power system signals of various frequencies. We find that the MPC algorithm incorporating the full delay distribution, versus only the mean delay, reduces the average tracking error 39%. Also, incorporating individual TCL models, identified on-line, within the state estimator versus only using a TCL aggregation model reduces the average estimation error 19%.