Common Linear Model Research Articles

Network Constrained Fast Decoupled Load Flow Solutions Using Linear Programming

The fast decoupled load flow method is now well established as a reliable means for evaluating the steady-state conditions of large scale power system networks. Although the method incorporates a limited number of constraints on control variables, noteably those on transformer taps and generator VAr injections, it is frequently necessary for planning purposes to perform repeated studies in order to achieve satisfactory voltage profiles and active and reactive power flows. The imposition of constraints on bus voltages and power flows, through control of power and reactive power injections and transformer and phase shifter tap changes, considerably reduces the number of studies required. This paper describes how the equations from the fast decoupled load flow method can be used in conjunction with linear models of control devices to apply the necessary constraints through linear programming. The method is to solve iteratively the fast decoupled load flow problem and a decomposed incremental 1 inear programming problem until all constraints are satisfied. Such are the accuracies of the incremental linear models that normally only one iteration is required, and overall computing effort is minimized by choosing common linear network models for load flow and linear programming.

On estimation of diagonal covariance matrices by minque

Let Y = Xβ + ϵ be a general linear statistical model with co-variance matrix of the form , where δi, is a known diagonal matrix and σi is an unknown scalar. This form includes the I common linear regression model with uncorrelated observations, with or without replications; the random coefficients model of Hildreth and Houck (1968); and numerous mixed effects models after a suit-able diagonalizing transformation. The MINQUE of is derived by a simple, intuitive approach in which a weighted least squares criterion is applied to the vector of squared residuals. In addition to providing a different perspective, this approach has the advantage that estimation can be restricted to the known parameter space. The restricted estimates are always nonnegative, but still equal to the MINQUE values when the latter are also nonnegative. A hypothetical example is simulated in which the restricted and unrestricted estimates are compared. The practical significance for an iterative feedback procedure is discussed, and som...