Non-intrusive load monitoring algorithm based on features of V–I trajectory

• Published in: Electric power systems research Vol. 157; pp. 134 - 144
• Main Authors: Wang, A. Longjun, Chen, B. Xiaomin, Wang, C. Gang, Hua, D.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.04.2018; Elsevier Science Ltd
• Subjects: Algorithms; Automatic meter reading; Electric appliances; Energy consumption; Energy management; Feature extraction; Interpolation; Load disaggregation; Load signatures; Monitoring; Monitoring systems; Multi-classification; Non-intrusive load monitoring; Power consumption; Smart metering; Studies; Support vector machines; Trajectory analysis; V–I trajectory; Non-intrusive load monitoring; Multi-classification; Smart metering; Load signatures; V–I trajectory; Load disaggregation
• ISSN: 0378-7796; 1873-2046
• DOI: 10.1016/j.epsr.2017.12.012

•Proposing a V-I trajectory extraction approach based on the steady-state data before and after an event.•The method of quantifying the V-I trajectory feature is proposed and the number of trajectory features is expanded.•C-SVC multi-classification method is applied for load recognition.•The algorithm is tested with both the REDD database and the laboratory data. Non-intrusive load monitoring (NILM) can monitor the status of electrical appliances on-line and provide detailed power consumption data, which is the basis for customers to perform energy usage analyses and electricity management. The voltage–current (V–I) trajectory can be used as a load signature to represent the electrical characteristics of appliances with different statuses. Therefore, this paper proposes an NILM algorithm based on features of the V–I trajectory. The variation in the overall apparent power was used as the criterion of event detection, and the delta of the V–I trajectory was extracted by smoothing and interpolation. Then, ten V–I trajectory features were quantified based on physical significance, which accurately represented those appliances that had multiple built-in modes with distinct power consumption profiles. Finally, the support vector machine multi-classification algorithm was employed for load recognition. We tested the proposed algorithm on both the REDD database and laboratory data. The numerical results demonstrate that the algorithm has higher accuracy than the algorithm using other load features.