Dynamic State Estimation: An improved approach for Wide Area Measurement Systems

Abstract

Most of power system components (generators, transmission lines, etc.) operate at their maximum capacities. Also because of constantly increasing generation and loading, transmission capacity must be increased proportionally to relieve the transmission system from stress. Emphasis over continuous and uninteruppted power supply leads to transmission lines working close their maximum loadability limits. Under such conditions a small disturbance may lead to severe consequences like cascade tripping and blackout, if proper decisions are not taken. In such a scenario operators need reliable information to operate. Rather than the conventionally available measurements there must be a more reliable source to obtain variables that truly describe the operation of the power system.

State Estimation allows calculation of such variables even when measurements are missing, inaccurate or corrupted by noise. Accurate state estimation allows an adequate margin to operating limits in case a fault occurs. State Estimation process forms the backbone for other power system operations, be it optimal power flow or contingency analysis. Faster state estimation of the system is important for better visualization and planning of the corrective actions. In its present condition, state estimation is static in nature. Static State Estimation assumes system to be at steady-state and does not incorporate system dynamics. However, this is not a real-time power system scenario. Thus, Dynamic State Estimation has been of great interest and importance. It aims at inclusion of dynamics of power system into the process of state estimation for better and reliable operation. The use of dynamic states like generator angle and speeds help incorporate the true system dynamics.

This thesis includes basics about various dynamic state estimation technique and focuses mainly on kalman filter based techniques. Extended and Unscented kalman filter techniques; which are kalman filter variants applicable to non-linear systems are studied in detail and comparison is provided. To demonstrate dynamic state estimation, a 3 generator 9-bus system is simulated. State estimation under various disturbance conditions are presented.