Mehmet K. Celik

Identifying the unknown circuit breaker statuses in power networks

This paper describes an approach by which the circuit breaker status errors can be detected and identified in the presence of analog measurement errors. This is accomplished by using the least absolute value (LAV) state estimation method and applying the previously suggested two stage estimation approach by A. Monticelli (see ibid., vol.8, no.3, p.1143-9, 1993). The ability of the LAV estimators to reject inconsistent measurements, is exploited in order to differentiate between circuit breaker status and analog measurement errors. The first stage of estimation uses a bus level network model as in conventional LAV estimators. Results of stage 1 are used to draw a set of suspect buses whose substation configurations may be erroneous. In the second stage, the identified buses are modeled in detail using the bus sections and the circuit breaker models while keeping the bus level network models for the rest of the system. The LAV estimation is repealed for the expanded system model and any remaining significant normalized residuals are flagged as bad analog measurements, while the correct topology is determined based on the estimated flows through the modeled circuit breakers in the substations. The proposed approach is implemented and tested. Simulation results for cases involving circuit breaker status and/or analog measurement errors are provided.

A robust WLAV state estimator using transformations

The authors present a robust weighted least absolute value (WLAV) power system state estimator which remains insensitive to bad measurements even when these are associated with leverage points. 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 is 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. >

A fast algorithm for the weighted least absolute value state estimation (for power systems)

The weighted least absolute value minimization method is used to solve the power system state estimation problem. The problem can be shown to be equivalent to a linear programming program and hence can be solved using the well-known Simplex method. However, a direct adaptation of the Simplex method results in an inefficient code. A fast algorithm originally suggested by I. Barrodale and F.D.K. Roberts (1973) is further improved and applied to the power system state estimation problem. The use of the algorithm resulted in considerable savings in computation times, as demonstrated by the numerical examples given for various test systems. >

A fast algorithm for the weighted least absolute value state estimation

The weighted least absolute value minimization method is used to solve the power system state estimation problem (PSSEP). In this paper, a fast algorithm originally suggested by Barrodale and Roberts is further improved and applied to the PSSEP. The use of the developed algorithm resulted in considerable savings in computation times as demonstrated by the numerical examples given for various test systems

Use of scaling in WLAV estimation of power system states

The authors present the results of employing a particular scaling strategy in weighted least absolute value (WLAV) state estimation of various test systems of sizes as large as 1500 buses. Apart from the effects of scaling on the leverage points and the robustness of the estimator, effects on computational speed and the dependence of these effects on the measurement types are also investigated. 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. >

Observations regarding compactness in the ∂-Neumann problem

In a first part, we show that compactness of the ∂-Neumann operator is independent of the metric, among metrics smooth on the closure of the domain. This is analogous to subellipticity, but in contrast to global regularity. Our methods also give a new proof of the independence of subellipticity. In a second part, we look at the ideal of compactness multipliers. The zero set of this ideal may be viewed as the obstruction to compactness. We determine this zero set in the case of convex domains and in the case of certain Hartogs domains in ℂ2.

Implementation of the power system weighted least absolute value state estimator using an improved l/sub 1/ linear approximation algorithm

Power system states can be estimated using the weighted least absolute value (WLAV) estimation technique. The estimation procedure is carried out iteratively by solving an l/sub 1/-linear approximation problem at each iteration. While the approximation problem can be solved by the standard simplex method, the special structure of the system equations makes further improvements in the algorithm possible. Such a procedure is developed using an algorithm originally suggested by L. Barrodale and F.D.K. Roberts (1973). The adaptation of this algorithm to the solution of the power system states in a WLAV estimator is carried out, and the problem size is considerably reduced, exploiting the characteristics of the measurement equations. Implementation details of the developed program together with some numerical results of running it on realistic size power systems are presented.<<ETX>>