Thermostatically Controlled Load Data Reconstruction Based on Improved Similar-Day Algorithm

• Published in: 2025 IEEE 3rd International Conference on Power Science and Technology (ICPST) pp. 1056 - 1061
• Main Authors: Lin, Qiao, Ding, Li, Li, Xin, Yu, Zhen-Wei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16.05.2025
• Subjects: Accuracy; data reconstruction; Demand response; Interpolation; Measurement; Meteorology; Power system reliability; Reliability; Root mean square; similar-day algorithm; Smart grids; Thermostatically controlled load; Time series analysis
• DOI: 10.1109/ICPST65050.2025.11088974

Thermostatically controlled loads (TCLs) are essential demand response resources with inherent flexibility, offering significant potential for optimizing power system operation and enhancing energy efficiency. Accurate reconstruction of TCL data is crucial for energy management, and evaluating demand response potential. However, missing or incomplete data caused by sensor failures, communication issues, and storage limitations pose challenges to practical TCL analysis. To address these issues, this study proposes an improved similar-day algorithm (ISDA) for reconstructing unavailable TCL data. The ISDA approach refines the selection of reference days by incorporating multiple similarity metrics, including time-of-day consumption patterns, weather dependencies, and seasonal variations, to enhance reconstruction accuracy. The ISDA and the benchmark method employ mean absolute percentage error and normalized root mean square error as primary accuracy metrics to evaluate its performance. Experimental validation on real-world TCL datasets demonstrates the superior performance of ISDA in preserving temporal correlations and accurately reconstructing missing data. These findings improve the quality of TCL data for demand response applications and provide valuable insights for optimizing energy management strategies in smart grids.