An expert system based framework for an incipient failure detection and predictive maintenance system

Abstract

Traditional preventive maintenance operations are being abandoned and electric utilities are becoming more failure driven due to the financial constraints being placed on them. When some distribution equipment begins to deteriorate, intermittent incipient faults persist in the system from as little as several days to several months. The failure of equipment in power distribution systems can have a direct or indirect impact on the reliable delivery of quality power. Also, certain failures can result in loss of service. There is great interest in the utility industry for low-cost, automated, real-time approaches which can detect distribution incipient faults and locate their source. This paper discusses an expert system based incipient failure detection and predictive maintenance (FDPM) system being developed for application in distribution systems. The FDPM system includes an expert system engine, a knowledge base, mathematical and neural network models of aging of distribution equipment, historical measurements databases, a distribution state estimator, a fault and disturbance event locator and a distribution system interconnection map. The FDPM system detects incipient disturbances, classifies the type of disturbance, and locates the source of the incipient behavior. If the source is one of the components under observation by the FDPM system, it assesses the integrity of the distribution system component and predicts maintenance needs.