Date of Award

Spring 5-15-2022

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical Engineering and Computer Science

Advisor(s)

Eftekharnejad, Sara

Subject Categories

Electrical and Computer Engineering | Engineering

Abstract

Over the last decades, great efforts and investments have been made to increase the integration level of renewable energy resources in power grids. The New York State has set the goal to achieve 70% renewable generations by 2030, and realize carbon neutrality by 2040 eventually. However, the increased level of uncertainty brought about by renewables makes it more challenging to maintain stable and robust power grid operation. In addition to renewable energy resources, the ever-increasing number of electric vehicles and active loads have further increased the uncertainties in power systems. All these factors challenge the way the power grids are operated, and thus ask for new solutions to maintain stable and reliable grids. To meet the emerging requirements, advanced metering infrastructures are being integrated into power grids that transform traditional grids into '' smart grids". One example is the widely deployed phasor measurement units (PMUs), which enable generating time-synchronized measurements with high sampling frequency, and pave a new path to realize real-time monitoring and control in power grids. However,the massive data generated by PMUs raises the questions of how to efficiently utilize the obtained measurements to understand and control the present system. Additionally, to meet the communication requirements between the advanced meters, the connectivity of the cyber layer has become more sophisticated, and thus is exposed to more cyber-attacks than before. Therefore, to enhance the grid reliability with PMUs, robust and efficient grid monitoring and control methods are required.

This dissertation focuses on three important aspects of improving grid reliability with PMUs: (1) power system event detection; (2) impact assessment regarding both steady-state and transient stability; and (3) impact mitigation. In this dissertation, a comprehensive introduction of PMUs in the wide-area monitoring system, and comparisons with the existing supervisory control and data acquisition (SCADA) systems are presented first. Next, a data-driven event detection method is developed for efficient event detection with PMU measurements. A text mining approach is utilized to extract event oscillation patterns and determine event types. To ensure the integrity of the received data, the developed detection method is further designed to identify the fake events, and thus is robust against cyber-threat. Once a real event is detected, it is critical to promptly understand the consequences of the event in both steady and dynamic states. Sometimes, a single system event, e.g., a transmission line fault, may cause subsequent failures that lead to a cascading failure in the grid. In the worst case, these failures can result in large-scale blackouts. To assess the risk of an event in steady state, a probabilistic cascading failure model is developed. With the real-time phasor measurements, the failure probability of each system component at a specific operating condition can be predicted. In terms of the dynamic state, a failure of a system component may cause generators to lose synchronism, which will damage the power plant and lead to a blackout. To predict the transient stability after an event, a predictive online transient stability assessment (TSA) tool is developed in this dissertation. With only one sample of the PMU voltage measurements, the status of the transient stability can be predicted within cycles. In addition to the impact detection and assessment, it is also critical to identify proper mitigations to alleviate the failures. In this dissertation, a data-driven model predictive control strategy is developed. As a parameter-based system model is vulnerable to topology errors, a data-driven model is developed to mimic the grid behavior. Rather than utilizing the system parameters to construct the grid model, the data-driven model only leverages the received phasor measurements to determine proper corrective actions. Furthermore, to be robust against cyber-attacks, a check-point protocol, where past stored trustworthy data can be used to amend the attacked data, is utilized. The overall objective of this dissertation is to efficiently utilize advanced PMUs to detect, assess, and mitigate system failure, and help improve grid reliability.

Access

Open Access

Download

Share

COinS