A deep learning approach for detecting distributed generation in residential customers. (March 2023)

• Record Type: Journal Article
• Authors: Al Khafaf, Nameer; Wang, Jia; Jalili, Mahdi; Sokolowski, Peter
• Is Part Of: Sustainable energy, grids and networks. Volume 33(2023)
• Publication Date: 2023-03
• Subjects: Smart meter -- Classification -- Deep learning -- Knowledge discovery -- Energy consumption; Renewable energy sources -- Periodicals; Smart power grids -- Periodicals; Electric power systems -- Periodicals; 333.794
• DOI: 10.1016/j.segan.2022.100966
• Languages: English
• ISSNs: 2352-4677

Abstract

Abstract: Penetration levels of distributed generation is expected to increase significantly worldwide in the coming years resulting in several emerging issues for the distribution industry. In countries where there is no strong public policy or legal framework to support distribution network operators with the integration of distributed generation to the grid, energy consumers may connect their generators to the grid without notifying distribution operators. This results in a significant impact on the distribution network if left undetected. The energy consumption from smart meters have been widely recognized as a key enabler for delivering a range of benefits to electricity industry. In this paper, two deep learning models are trained for two purposes; ( i ) to detect residential customers with distributed generation, and ( ii ) to identify the exact date or range of dates of when distributed generator came online based on energy consumption only. These two models can be integrated as part of the distribution operators' tools to update customers' records. The results show that deep learning can detect distributed generation with an accuracy of > 98%. The research is based on real energy consumption datasets provided by an Australian distribution network operator.