Power System State Estimation In Large-Scale Networks

Power system state estimation constitutes one of the critical functions that are
executed at the control centers. Its optimal performance is required in order to operate
the power system in a safe, secure and economic manner. State estimators (SE)
process the available measurements by taking into account the information about the
network model and parameters. The quality of estimated results will depend on the
measurements, the assumed network model and its parameters. Hence. SE requires to
use various techniques to ensure validity of the results and to detect and identify
sources of errors. The Weighted Least Squares (WLS) method is the most popular
technique of SE. This thesis provides solutions to enhance the WLS algorithm in
order to increase the performance of SE. The gain and the Jacobian matrices
associated with the basic algorithm require large storage and have to be evaluated at
every iteration, resulting in more computation time. The elements of the SE Jacobian
matrix are processed one-by-one based on the available measurements, and the
Jacobian matrix, H is updated suitably, avoiding all the power flow equations. thus
simplifying the development of the Jacobian. The results obtained proved that the
suggested method takes lesser computational time compared with the available NRSE
method, particularly when the size of the network becomes larger. The uncertainty in
analog measurements could occur in a real time system. Thus, the higher weighting
factor or wrongly assigned weighting factor to the measurement could lead to flag the
measurements as bad. This thesis describes a pre-screening process to identify the bad
measurements and the measurement weights before performing the WLS estimation
technique employed in SE. The autoregressive (AR) techniques. Burg and Modified
Covariance (MC), are used to predict the data and at the same time filtering the
logical weighting factors that have been assigned to the identified bad measurements.
The results show that AR methods managed to accurately predict the data and filter
the weigthage factors for the bad measurements. Also the WLS algorithm is modified
to include Unified Power Flow Controller (UPFC) parameters. The developed
methods are successfully tested on IEEE standard systems and the Sabah Electricy
Sdn. Bhd. (SESB) system without and with UPFC. The developed program is suitable
either to estimate the UPFC controller parameters or to estimate these parameter
values in order to achieve the given control specifications in addition to the power
system state variables.