Multi-Time-Scale Modeling and Parameter Estimation of TCLs for Smoothing Out Wind Power Generation Variability

Abstract

Thermostatically controlled loads (TCLs) have demonstrated their potentials in demand response. One of the key challenges for TCLs to be integrated into the system-level operation is building a compact aggregated model, in which the TCL primary behaviors are accurately captured. In this paper, TCLs are aggregated as a virtual generator and two batteries according to their different compressor types and control methods for smoothing out multi-time-scale variability of wind power generation. This will bring system operator great convenience to manage TCLs and conventional components when the system-level decisions are made. Accordingly, accurate parameters of virtual generator and batteries are critical to effectively coordinate TCLs with other resources in the system operation. However, it tends to be difficult to obtain such aggregated parameters as a result of insufficient data for each TCL. To address this problem, high-dimensional model representation (HDMR) is introduced to estimate the aggregated parameters of virtual generator and batteries using the probability distribution of TCL parameters. A numerical simulation study demonstrates that aggregated parameters of virtual generator and batteries can be accurately estimated by HDMR. And virtual generator and batteries are able to follow actual behaviors of TCL populations in power system operations.