Modern power systems are complex in nature. Accurate knowledge and estimation of the electro-mechanical modes in the power system are of great importance since a system-wide outage can be caused by one single unstable mode of oscillation. Hence, advanced mathematical tools and signal processing methods are necessary to estimate the electro-mechanical modes of the power system. There have been several incidences where system breakups and power outages happened due to an undamped mode and all of these events exhibited sustained low-frequency inter-area oscillations which typically are in the range of 0.15Hz to 1Hz. Model-based frequency studies have been established for a long time. However, model-based studies cannot anticipate every kind of event in real-time because the power system is nonlinear, time-varying and difficult to represent in its full higher-order dynamics. Besides, the power grid is stochastic with the advent of renewables and changing load dynamics, which means that it is expected to be excited by random signals all the time; most of them come from random load changes and noises. Due to these factors, Recently, measurement-based power grid mode estimation has attracted great attention.
Through this research work, first a comparison of various measurement-based electro-mechanical oscillation mode detection methods are studied. Several techniques can identify oscillatory modes from ring-down data or ambient data obtained from measurement. Generally, ring-down events occur during the sudden changes in grid operation due to faults, generator outages, or controller operations of the generator. Ambient data changes are rather slow varying close to the steady-state operation of the grid due to small changes in the load or such events. Among the measurement-based techniques, Prony analysis, Eigenvalue Realization Algorithm (ERA) and Matrix Pencil (MP) methods are noteworthy. These methods have successfully been used to determine low-frequency oscillatory modes from measurement data. Recently Subspace Identification (SSI) methods have become popular as they are robust to variations, and can be represented in state-space form thus making it easier for designing time-domain control approaches. As a result, an online wide-area direct coordinated control architecture for power grid transient stability enhancement based on subspace identification and Lyapunov energy functions has been studied. Also, a novel combined deterministic-stochastic online measurement-based identification framework using subspace theory is introduced. The novelty of the design using a fully recursive algorithm and the effectiveness of combined treatment are further discussed. Third, an approach for power system oscillatory mode estimation and classification based on the proposed method are discussed. For controlling electro-mechanical oscillations of power system effectively, identifying the location of oscillation source is very important. Identifying the location of oscillatory mode based on Lyapunov energy functions is also being explored in this work. Identifying oscillation source location is as important as identifying the mode itself. The most commonly used method for identifying oscillation source location based on measurement data is dissipated energy flow (DEF) method. In this work, a new method is proposed and performance against dissipated energy flow method is being explored. Finally this work explores grid following and grid forming control architecture and their usage for integrating distributed energy resources (DERs) into existing grid technology. Integrating different distributed energy resources into current grid can cause some stability issue. This work explains grid following and grid forming technologies and use of identification method to observe low frequency oscillation for DER connection.