Inrush transient generation and line impedance estimation

Abstract

An inrush transient contains extensive information that permits load identification, condition monitoring, and line impedance estimation. A power system monitor's (PSM) ability to identify a load based on its inrush behavior depends on the training exemplars used to create and optimize the load identification algorithm. This work discusses the use a phase-controlled switch that can be used in situ to integrate the effects of source and line impedance into the inrush data, and to generate transients at controllable turn-on phase angles relative to the voltage line-cycle. The resulting exemplars are more realistic than those generated with conventional techniques such as testing with an AC power supply. The control over angle also enables efficient investigation of a load's transient variability space. Testing loads in fault conditions expands the variability space, allowing load identification algorithms to correctly identify faulty loads and perform diagnostics. The large, high-frequency current that inrush transients inject into the line provides excellent excitation for line impedance estimation. Previous switching based approaches focus on fitting the line impedance to a model, i.e. parametric impedance estimation. This thesis extends previous work by providing the current excitation with common electrical loads rather than using capacitors, inductors, and short circuits. Non-parametric impedance estimation is also demonstrated. Inrush transients, and other transients generated by switching the load on and off rapidly, generate current with wide-bandwidth spectral content to replace previously used sinusoidal injection sweeps.