Electric power lines are subject to disturbances such as severe weather conditions, contact with animals, falling trees, human accidents, equipment malfunction, etc. These catastrophic events can trigger an electrical fault, which results in momentary or sustained interruptions. The interruptions cost U.S economy around $114 billion annually. Almost 80% of fault-related interruptions experienced by customers are due to fault incidents on the distribution system (DS). If the fault location is not identified accurately, power outages can last for a very long time and impact the economy severely. Fault Location Identification (FLI) methods of transmission systems are not easily applicable to DS because their accuracy is affected by fault types, fault resistance, unbalanced loads, and distributed generation (DG). FLI is also a challenging problem for electric utilities due to the scarcity of data points. However, due to the recent deployment of advanced metering infrastructure (AMI) and synchrophasor technology (ST), data availability is growing continuously. The widespread popularity of µ phasor measurement units and smart meters (SMs) in the context of smart grid technology has opened many doors to the use of data analytics and state-of-the-art algorithms for FLI in distribution networks (DN). Distribution-level PMUs, commonly known as µPMUs provide GPS-synchronized measurements of three-phase voltage and current phasors at a very high resolution. Similarly, SMs provide energy and power readings as fast as a one-second interval. Utilizing the measurements provided by an optimum number of µPMUs and SMs, this thesis presents and validates FLI techniques for a modern DN. Two approaches are proposed for FLI - a state estimation based approach and a machine intelligence-based approach. Using the state estimation method, FLI is performed using real-time data from simulated phasor measurement units (PMU), placed in the DN. State Estimation needs the fault currents of the generators and voltage measurements of an optimal number of nodes to perform the FLI algorithm. The method was validated using the IEEE 37 node test feeder, which mimics the true behavior of an operating distribution grid with DG. PMUs are placed on the real-time model of the system. The real-time model was implemented on a digital real-time simulator (DRTS), which streams phasor data over the Internet using C37.118 protocol. OpenPDC is used to collect real-time PMU data coming from real-time simulator. Microsoft SQL is used as a database management server (DMS) to store data coming from OpenPDC. In the last step of the FL process, data stored in OpenPDC is fed into a Matlab based FL identification algorithm to locate the fault. Both balanced and unbalanced fault types are applied to different nodes, and an accurate estimation of the FL (over 90% of the cases) is achieved. The machine intelligence-based FLI method is proposed using artificial neural networks (ANN’s) and ensemble learning using advanced metering infrastructure (AMI). This method also describes the development of a testbed for real-time testing of the proposed approach. The testbed consists of a simulated power system model running on a DRTS and AMI. The core parts of AMI are smart meters (SMs), a communication network (developed using DNP3 protocol over TCP/IP), a data concentrator (DC), and a utility operations center (UOC). Event-driven data from the SMs are collected in DC and then fed to the UOC for being used as inputs for the novel FLI algorithm. Based on the data received, the algorithm can classify the type of fault and locate it with high accuracy. Both balanced and unbalanced fault types are tested on different nodes and lines throughout a distribution network modeled in offline and on the DRTS. A comprehensive sensitivity analysis is performed to validate the effectiveness of the proposed method by varying fault resistance, loading conditions, and noise level. Classification accuracy of over 99% is achieved when classifying all fault types, and more than 95% accuracy is achieved when identifying the location of the fault.
