A Neural Network-Based Classifier for Identifying and Locating Neutral Wire Breaks in Low Voltage Distribution Networks

The breakage of the neutral conductor in low voltage distribution networks is a major concern for distribution companies. This breakage causes significant voltage deviations that can damage the connected equipment as well as jeopardizing people. The detection and localization of the breakage is a major challenge as it does not always manifest in the same way. This work presents a methodology based on artificial intelligence for the detection and localization of neutral conductor breaks in distribution networks. Two neural networks are trained in attempt to solve each of the challenges. For this purpose, measurements commonly taken by smart meters such as power and nodal voltages are used. The methodology is evaluated in simulation exhibiting a good performance.