An Active Sensing Method of Power System Load Network Based on Sweep Frequency Response Analysis

• Published in: IEEE transactions on instrumentation and measurement Vol. 72; pp. 1 - 14
• Main Authors: Lu, Delong, Tong, Chong, Wu, Zhijian, Huang, Xinyi, Wang, Xinhao, Zhou, Lei, Xu, Qingshan
• Format: Journal Article
• Language: English
• Published: New York IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Active sensing; Algorithms; Clusters; Demand analysis; Electrical loads; Electrical products; Energy consumption; energy disaggregation; Feature extraction; Frequency analysis; Frequency response; Home appliances; Household appliances; Impedance; Input impedance; load location; Load modeling; load network; Mathematical models; nonintrusive load monitoring (NILM); Parameter identification; Power lines; same load; Sensors; Smart meters; Sweep frequency; sweep frequency response analysis (SFRA); XGBoost
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2023.3329167

Apart from demand-side energy disaggregation, obtaining a more fine-grained energy consumption sensitivity is a remarkable improvement toward low-carbon. Aiming at the problems of load identification and location with the same type, parameter, and working conditions, which are difficult to be realized by the existing nonintrusive load monitoring (NILM) methods, a novel active sensing method of power system load network based on sweep frequency response analysis (SFRA) is proposed. First, the user-side load network model is introduced into the household appliance load cluster behind smart meters, and the power line that transmits electric energy is regarded as a part of the load cluster to form the user-side load network model. Second, based on the difference between load impedance and input impedance under the distributed parameter model, the spectrum characteristics of input impedance are obtained by SFRA, so as to realize the identification and location of the same load. Finally, the fitting and filling of load network matrix parameters are realized based on the XGBoost algorithm, and the positive results obtained seem to confirm the applicability of the proposed method. The identification signal changes from passive "load signature" to active "impedance sensing," which solves the problems of the same load identification and location that are difficult to deal with by the existing methods.