Power system state estimation and probabilistic loadflow analysis.

This thesis investigates the treatment of aposteriori and apriori uncertainties in power system planning and operation. Aposteriori uncertainty is treated by power system state estimators. A survey of existing techniques and their limitations is described. A method is presented that improves the speed of weighted least squares state estimation by modifying the structure of injection measurements to give a very sparse information matrix, the matrix to be inverted. Used with fast decoupling, this approach yields a very fast on-line state estimator, capable of handling all types of measurements. Bad data detection and identification techniques are reviewed and an improvement based on "mathematical" bad data removal is presented. The inclusion of h.v.d.c. links into a.c. state estimation is considered. Decoupling and geographic partitioning of the multi a.c. -h.v.d.c. state estimator are shown to cause little degradation in the estimates, and a method of accurately representing commutation overlap angle is outlined. Availability analysis in state estimator operation and design is considered, and applied to optimal meter placement design. The feasibility of hierarchical central-electrical, local-dynamic hydroturbine and canal state estimation, based on a linearized Kalman filter, is investigated. Apriori uncertainty in long-term future planning studies involving expected nodal generation and loads can be included in stochastic loadflows. A method is presented which enables the stochastic loadflow, which handles only gaussian statistics, to handle non-gaussian probability distributions via gaussian sum approximations. H.v.d.c. links are also included in a.c. stochastic loadflows, using both correlated and uncorrelated data