Distribution Systems Fault Location Identification Using Mixed Datasets

This paper proposes a method for identifying fault locations in active distribution networks equipped with inverter-interfaced distributed generations (IIDGs). The proposed method is capable of utilizing all available mixed datasets of modern distribution networks to improve the performance and precision of the fault location identification results. The dataset may include analog/oscillography data and discrete/status data. A causal model for the faulted system based on probabilistic Petri-Nets and backward reachability analysis is developed that utilizes the collected discrete/status data to determine the faulted section of the system. The model accounts for possible failures of protective devices and false notifications from fault indicators. An enhanced impedance-based fault location identification method is proposed that utilizes analog data from micro-PMUs and/or legacy measuring devices to estimate the exact location of the fault within the identified faulted section. It explicitly accounts for the uncertainties in the collected data. The proposed method's performance is showcased by simulating the IEEE 34-node distribution test system.