Enhanced power system resiliency to high-impact, low-frequency events with emphasis on geomagnetic disturbances

Power system resiliency;Corrective control;Cyber-physical attacks;Data analysis;Geomagnetic disturbances;State estimation

Kazerooni, Maryam

关键词: Power system resiliency;    Corrective control;    Cyber-physical attacks;    Data analysis;    Geomagnetic disturbances;    State estimation;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/95350/KAZEROONI-DISSERTATION-2016.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Various reliability procedures have been developed to protect the power systems against common reliability issues that threaten the grid frequently. However, these procedures are unlikely to be sufficient for high-impact low-frequency (HILF) events. This thesis proposes several techniques to enhance resiliency with respect to HILF events. In particular, we focus on cyber-physical attacks and geomagnetic disturbances (GMDs). Corrective control through generation redispatch is proposed to protect the system from cyber-physical attacks. A modification of the optimal power flow (OPF) is proposed which optimizes the system resiliency instead of the generation cost. For larger systems, the burden of solving the resilience-oriented OPF is reduced through a fast greedy algorithm which utilizes proper heuristics to narrow the search space. Moreover, an effective line switching algorithm is developed to minimize the GMD impact for large-scale power systems. The algorithm uses linear sensitivity analysis to find the best switching strategy and minimizes the GIC-saturated reactive power loss. The resiliency may be improved through power system monitoring and situational awareness.Power system data is growing rapidly with the everyday installation of different types of sensors throughout the network. In this thesis, various data analytics tools are proposed to effectively employ the sensor data for enhancing resiliency. In particular, we focus on the application of real data analysis to improve the GMD models. We identify common challenges in dealing with real data and develop effective tools to tackle them. A frequent issue with model validation is that for a real system, the parameters of the model to be validated may be inaccurate or even unavailable. To handle this, two approaches are proposed. The first approach is to develop a validation framework which is independent of the model parameters and completely relies on the measurements. Although this technique successfully handles the system uncertainties and offers a robust validation tool, it does not provide the ability to utilize the available network parameters. Sometimes, the network parameters are partially available with some degree of accuracy and it is desired to take advantage of this additional information. The second validation framework provides this capability by first modifying the model to account for the missing or inaccurate parameters. Then a suitable validation framework is built upon that model. Another common issue that is widely encountered in data analysis techniques is incomplete data when part of the required data is missing or is invalid. Examples of missing data are provided through real case studies, and advanced imputation tools are developed to handle them.