Time and order recursive estimation of power system electromechanical modes using synchro-phasors

The electrical power system is a critical infrastructure. Power system blackouts have deep social and economic ill-impacts. Deregulation of the electricity market, decrease in the rate of development of new energy infrastructure, higher penetration of non-committable renewable generators and ever increasing energy demand have forced the system operators to work the system close its stability limits and therefore closer to such eventualities. Consequently, it has become imperative that the stability of the system be determined in real-time.;The steady state stability of the system is often described in terms of the frequency and damping of the electromechanical modes of the system. Recent advances in synchrophasor technology have allowed researchers and operators to supplement the traditional model driven calculations of the modes with measurement based estimation techniques. These estimates are sensitive to user choices such as the number of parameters to be used.;This dissertation presents a real-time algorithm with a modular structure for the estimation of the modes using ambient measurements from the grid. The modularity of the method alleviates the requirement of fixing the number of parameters by providing simultaneous estimates for all choices of the length of the prediction filter. Furthermore, a real-time algorithm may be troubled by bad data. To that end, the dissertation modifies the said algorithm to cope with such situations.;The confidence in mode estimates is expressed as an estimate of standard deviation for both the frequency and damping estimates. Two methods are presented -- an analytical method and a quasi-numerical method. This dissertation also examines recursive whiteness testing of the prediction residuals as a means of model validation. The whiteness testing is useful in suggesting a reasonable model order.;Robustness, order recursiveness and real-time qualities of the algorithm, together with methods for estimating standard deviations in the mode estimates and recursive whiteness testing, provide a good alternative to the traditional estimation techniques. The applicability of these methods is verified using simulated and measured synchro-phasor data from the western North American power system.