Efficient and robust weighted least absolute value state estimation for power systems /
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1991.
|
| Subjects: | |
| Online Access: | Link to OAKTrust copy |
| Abstract: | The objective of this dissertation was to devise algorithms to make the weighted least absolute value (WLAV) estimator fast enough for on-line applications and to improve its robustness. The third chapter addresses the computational efficiency of the WLAV estimator by reformulating the problem in a decomposed manner using the Dantzig-Wolfe decomposition principle. The detailed formulation of the state estimation problem by applying the rules of this decomposition method is given along with the implementational changes made to further speed-up the algorithm. The results of the numerical simulations carried out for various size test systems are also presented. The algorithm of the fourth chapter is based on a fast l1 approximation method first proposed by Barrodale and Roberts. Their method is further improved by taking into account the sparsity and the special structure of the power systems measurements. The use of the developed algorithm results in considerable savings in computation times as demonstrated by the numerical examples given for various test systems. Chapter V presents a robust WLAV estimator, which remains insensitive to bad measurements even when these are associated with leverage points. Existence of leverage points has been claimed to be the reason for the WLAV estimator failing to reject bad data in the measurements. Leverage points are evenly distributed in the factor space of multiple regression via linear transformations. These transformations represent a change of coordinates in the state space. The transformed system of measurement equations are then used to obtain the WLAV estimator for the system states. The transformation based WLAV estimator is shown to remain robust in the presence of leverage points by using some test cases. In chapter VI, the effectiveness of scaling in avoiding the leverage points in WLAV estimation of power system states is investigated. Scaling has been widely used in linear programming problems for numerical stability and computational efficiency. Simulation results indicate that scaling not only helps to reduce the leveraging effects but also improves the computational performance by decreasing the required number of Simplex iterations. In summary, this dissertation investigates issues such as the applicability of the WLAV estimator to large-scale power systems and its robustness in the presence of multiple interacting bad data associated with leverage points. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. "Major subject: Electrical Engineering." |
| Physical Description: | xi, 147 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references. |