|Institution:||University of Illinois – Urbana-Champaign|
|Full text PDF:||http://hdl.handle.net/2142/72913|
This thesis proposes several online power system monitoring and operations tools that leverage real-time measurements with little or no reliance on models obtained offline. Instead of relying on offline models, which can be grossly inaccurate, as in conventional power system monitoring and operations, the proposed methods exploit high-speed synchronized measurements obtained from phasor measurement units (PMUs). These proposed methods infer up-to-date and pertinent information regarding the operating status of the power system, and they are desirable owing to their ability to adapt to the current system operating point and topology. The tools proposed in this thesis fall within two general categories: (i) those that are adaptive to changes in system operating point and topology, and (ii) those that detect and identify topology changes. With respect to the first category, a measurement-based method is developed to estimate power system linear sensitivities. The proposed method is used to estimate, in near real-time, up-to-date power system linear distribution factors and the power flow Jacobian matrix. Improvements and extensions to the general methodology are developed: (i) the number of required measurements is reduced in distribution factor estimation, (ii) distributed computation is described in the context of Jacobian matrix estimation, and (iii) estimated distribution factors resulting from the proposed measurement-based method are shown to be more ubiquitous than their conventional model-based counterparts. In the second category, a measurement-based method is proposed to detect and identify line outages using quickest change detection algorithms. The proposed scheme is evaluated by its average detection delay and the probability that the wrong line outage is identified. In order provide support for the tools proposed in this thesis, the results of all proposed measurement-based methods are compared with their model-based counterparts via examples and case studies.