Non-technical loss detection at unregistered points in power distribution systems

Abstract

In this paper a non-technical loss (NTL) diagnostic method is proposed which considers faults at unregistered points in the distribution network. The method relies on the concept of smart grid, where every consumer is equipped with smart meters. Existing line segments are split up in order to detect and isolate faults at arbitrary points between metered points. The goal is to adjust the power and distance of these nodes such that the measured voltage profile is obtained. This results in a non-linear optimization problem, which is approximated using a genetic algorithm. Based on conservation of energy a local evaluation method is worked out. This is used to reduce the enormous search space of a fully extended network. The computational burden is further relaxed by showing the applicability of parallel computation for both diagnostic techniques.

In this paper a non-technical loss (NTL) diagnostic method is proposed which considers faults at unregistered points in the distribution network. The method relies on the concept of smart grid, where every consumer is equipped with smart meters. Existing line segments are split up in order to detect and isolate faults at arbitrary points between metered points. The goal is to adjust the power and distance of these nodes such that the measured voltage profile is obtained. This results in a non-linear optimization problem, which is approximated using a genetic algorithm. Based on conservation of energy a local evaluation method is worked out. This is used to reduce the enormous search space of a fully extended network. The computational burden is further relaxed by showing the applicability of parallel computation for both diagnostic techniques.