Novel Optimal PMU Placement Method for System Stability Enhancement and Deep Learning-Based PMU Security Model against Cyber Attacks

Abstract

Over the past years, new technologies have been integrated into the smart grid, adding new capabilities to the measurements and control of the transmission system. These technologies make the grid much more reliable, and minimize the possibility of wide spread power interruption. The placement of phasor measurement units (PMUs) in modern power systems provides improved monitoring and control characteristics of the entire electrical networks. These devices can accurately measure the phasor of an electric quantity and provide these measurements at a high sampling rate. The installation of a large number of PMUs is associated with relatively high cost and complicated communication infrastructure. As a result, the allocation of the PMU devices needs to be optimized to achieve observability of the entire system while minimizing the related cost. This thesis proposes a new method for solving the optimal PMU placement (OPP) problem using integer linear programming (ILP) such that the global optimal solution is guaranteed. As opposed to the reported studies in the literature, the proposed method considers the network parameters such as the series impedances and the shunt admittances of the transmission lines and transformers impedances in the formulation of the OPP problem. These parameters are found to have a crucial impact on the observability of the entire network. In addition, the proposed method seeks to achieve the maximum measurement redundancy while considering the impact of failures among individual PMU devices (N-1 contingency). Furthermore, the presented approach incorporates additional constraints which are associated with the control feedback signals of the supplementary damping controllers and thus enhance the small signal stability of a power system. The effectiveness of the proposed method is tested using different standard IEEE systems. The wide area damping controller uses remote signal as a feedback to provide proper damping to power system oscillations. This signal is usually obtained from the PMUs, because of their fast and accurate measurements. The benefits of PMUs introduce additional challenges of how to secure their data from being manipulated by cyber attackers, as the danger of cyber-attacks increases with the power grid reliance on communication and digital technologies. The cyberthreats can cause short-term power outage or even long-term system damage. Since most of modern power system applications relies on PMUs measurements, appropriate solutions should be implemented to detect these attacks and avoid any severe damage to the system. In this thesis, a new deep learning-based security model is proposed to detect PMU data manipulation attacks. The model not only detects attack existence, bus also it determines the attacked PMUs exactly. Unlike many approaches presented in the literature, the proposed model is able to capture the behavior of the system on both steady state and transient operation. The state-of-the-art deep learning techniques are used to train and test the model. The obtained results show superior performance when the proposed model is evaluated using IEEE 9-bus and 39-bus systems, as model succeeds to detect the cyber-attacks with very high accuracy.

Date of Award	Dec 2020
Original language	American English