| This dissertation has been focusing on automated analysis of power quality data and transmission line fault location. New contributions for automatic detection, classification and characterization of power quality disturbances, equipment sensitivity study during power quality disturbances, worst case capacitor switching transients determination as well as transmission line fault location have been achieved. Intelligent techniques such as the expert system, fuzzy logic, neural networks, genetic algorithms as well as advanced signal processing techniques like the wavelet analysis have been investigated and utilized for fulfilling the objectives.;For power quality disturbance detection and classification, we have established an innovative fuzzy expert system based decision making system, and extracted unique features utilizing both the Fourier and wavelet analysis.;To characterize power quality disturbances, various pertinent parameters for describing specific types of disturbances have been defined, and appropriate algorithms for deriving them have been developed.;We have developed a systematic approach for the equipment sensitivity study during various types of disturbances. The capabilities to consider all the common types of power quality disturbances and flexibly tune various waveform parameters, with the additional advantage of being able to utilize both the simulated and recorded data have been implemented.;A genetic algorithm based approach for determining the worst case capacitor switching transients has been developed. By iteratively changing the appropriate variables, running simulation, and processing and evaluating the simulated results, the genetic algorithm guides the searching process for an optimal solution.;To the end of fault location utilizing recorded data coming from recording devices like digital fault recorders sparsely located at various substations, we have developed a new "waveform matching" based approach. The fault location estimation is formulated as an optimization problem with the fault location and fault resistances being unknown variables. By posing faults, running simulation, and comparing the simulated and recorded waveforms, the fault location is determined as the one specified in the software that allows generating the simulated waveforms that best match the recorded ones. A simple yet effective genetic algorithm has been developed for guiding the searching process. |
