Doctor of Philosophy in Electrical and Computer Engineering (PhD)
Real-time Tools for Power System Optimization, Operation, and Control
In this thesis, we focus on two major real-time applications of modern synchrophasor-based wide-area measurement systems , i.e., transient stability assessment (TSA) and fault detection and identification (FDI). First, we develop a tool for real-time TSA based on automatic learning approaches. We use Classification and Regression Tree (CART) as the classification tool and Multivariate Adaptive Regression Splines (MARS) as the regression tool. To train and validate these tools in a practical setting, we conduct test cases on the full Western Electricity Coordinated Council system model, with emphasis on the BC Hydro (BCH) power system. While being mindful of practical field implementations of the proposed methods, our studies assume limited number of phasor measurement units (PMUs) installed, in accordance with existing infrastructure in the BCH system. The trained CART models are tested and show high accuracy rates, and thus, will be able to predict the transient stability issues of the system under study following different contingencies using the synchrophasors obtained from limited number of PMUs in the system. Also, the MARS models, which are proposed to be applied for TSA for the first time, show reasonable prediction accuracy rates. Next, we investigate the possibility of an accurate real-time FDI using synchrophasors received from PMUs installed at the two ends of a transmission line. We apply a new metric called goodness-of- fit (GoF), which is calculated over the time span of measurement and can quantify the credibility of the received synchrophasors. Then, we apply the data to an FDI method to show how accurate and credible the results are. The obtained results show a reasonable relation between the GoF metric, i.e., credibility of the measured sychrophasors, and the accuracy of the obtained results, validating the significance of the proposed method for real-time applications. As it is very rare to have a real power system with all buses and transmission lines equipped with PMUs, we also propose a wide-area real-time FDI approach using a linear observer. Through this wide-area approach, we demonstrate the effectiveness of the proposed method by accurately locating a fault in a small test system.