This dissertation presents a new technique for locating and diagnosing faults on electric power distribution feeders. Based on the integration of information available from disturbance recording devices with knowledge contained in a distribution feeder database, the proposed fault location scheme will make it possible to accurately identify the location of a fault upon its occurrence, thus speeding the restoration process. The developed fault location and diagnosis system can also be applied to investigate the causes of disturbances that may not result in a blown fuse and evaluate the performance of the protection scheme.; The proposed scheme involves three sequential processes: signal processing, fault location and fault diagnosis. The scheme first uses a signal processing algorithm to estimate the phasor quantities, identify the disturbance and extract the sequences of events from the digital voltage and current waveforms recorded at the substation. With the estimated voltage and current phasor quantities, a fault location algorithm is then applied to calculate the fault distance and search for possible fault locations. To deal with the uncertainties inherent in the system modeling and the phasor estimation, this algorithm is further extended to estimate a fault region for each calculated fault location based on probabilistic modeling and analysis. Since the distribution feeder is a radial network, multiple possibilities of fault locations could be obtained with measurements available only at the substation. To identify the actual fault location, a fault diagnosis algorithm is finally used to prune down and rank the multiple fault locations by integrating the available pieces of evidence through the use of a fuzzy pattern recognition technique. Testing of the developed fault location and diagnosis system using both simulated data and field data has verified the effectiveness of the proposed fault location and diagnosis scheme and demonstrated its potential for practical use.
展开▼
