Identification of optimal PMU locations and synchrophasor measurements based estimation of electromechanical modes in power system

Abstract

Phasor Measurement Units (PMU) are microprocessor-based devices used for real time monitoring of power system. PMU measures the voltage and current phasors of the buses in the power system and transfers the data to the control centre. This data is later used for various applications like stability assessment, adaptive relaying etc. However, due to the high installation cost of PMU and its related equipment, it is not economically viable to place the PMUs at every bus in the power system. Hence, this thesis aims to optimally place the PMUs maintaining complete observability of the power system during normal operating conditions and contingencies and develop algorithms using PMU data for identifying the poorly damped modes, which affect the small signal stability of the system. This thesis is divided into two sections. The first section deals with optimal placement of PMUs of different channel capacity along with maximization of measurement redundancy using an Integer programming technique. The effectiveness of the proposed model is tested on various IEEE test systems and a practical Indian power system for normal operating conditions as well as contingencies like single line outage and PMU outages. The second part of the thesis is focussed on developing algorithms for analyzing the poorly damped modes in power system low frequency oscillations utilizing synchrophasor measurements. The low frequency oscillations occurring in power system can be broadly classified into ambient and ringdown type oscillations. The ringdown oscillations occur when the power systems are subjected to large-magnitude disturbances, whereas, ambient type oscillations occur due to random small changes in load or generation. The persistence of these oscillations in power system will cause cascaded tripping leading to blackouts. Hence, to prevent such unwanted occurrences, it is essential to identify these oscillations at the earliest. As the characteristics of these oscillations are not similar, different algorithms are needed for the proper analysis of these oscillating modes. Algorithms based on Hankel’s Total Least Square (HTLS) and Estimation of Signal Parameters using Rotational Invariance Technique (ESPRIT) have been developed for analysing ringdown oscillations. The model order which is a prerequisite for the proper implementation of HTLS and ESPRIT algorithms, are obtained through an FFT based technique and EMO algorithm respectively. A Stochastic Subspace Identification (SSI) based method is developed to analyse ambient type of oscillations. To improve the robustness of the proposed SSI based method, a Stationary Wavelet Transform (SWT) based denoising method is used. Further, the model order of the signal is estimated using EMO algorithm. Subsequently, in order to avoid the computation of model order, an Empirical Wavelet Transform (EWT)- ESPRIT algorithm has been developed for identifying the poorly damped modes in power system. In this algorithm, the EWT is used for splitting the multi-component signal into mono components and ESPRIT algorithm is used for estimating the modal parameters of these mono components. This algorithm can analyse stationary as well as non stationary signals, which occur during a transient condition. The performance evaluation of these algorithms are conducted using synthetic signals with known modal parameters and real-time signals obtained from the PMUs installed in a practical system at different signal to noise ratios and PMU reporting rates. Results reveal that these algorithms perform better than the similar algorithms in literature.