Biologically Inspired Intelligent Fault Diagnosis for Power Distribution Systems

North Carolina State University, Raleigh NC

Investigators

Abstract

This project will investigate and develop a Biologically Inspired Intelligent Fault Management System using Artificial Immune System (AIS) technologies on top of a Neural Network - Fuzzy Logic (NN-FZ) structure to actively manage power distribution system faults, including diagnosis, prognosis, and data mining. The proposed approach can answer the challenges of discovering new fault diagnosis knowledge based on the dynamic operating environments, in addition to providing accurate fault diagnosis and prognosis. The NN-FZ technology will be used to aggregate information collected from sources such as SCADA system, circuit database, fault database, etc., to perform fault diagnosis and prognosis. Then an AIS algorithm will be used to guide the NN-FZ to learn, absorb, adapt, evolve and improve the fault diagnosis performance based on existing information such as the distribution system's network topology, operating conditions and weather conditions, and depending upon newly acquired information, such as new faults and operating conditions. This proposed system can data mine the existing outage database and explain heuristics about the fault diagnosis and prognosis process as preferred by operators and engineers. It can disseminate learned information to other power distribution centers, thereby preventing the reoccurrence of "learned" problems. This system would revolutionize the Fault Diagnosis process for power distribution systems, to significantly increase system reliability and reduce operation costs. The proposed activities and architectures are not only limited to power distribution system, but are also applicable to other industries such as communication networks and transportation system that are large scale nonlinear system with uncertain operating environments.