New global parameter estimation method for a transformer detailed model

Abstract

The RLCM parameters of the ladder network detailed model of a transformer winding are obtained using a new estimation method based on frequency response measurements on transformer terminals. The estimation problem is reformulated from electromagnetic formulations used in the model evaluation algorithm. A multi-objective global search for solutions is performed by means of a genetic algorithm in a space of reduced and fixed dimension. The coupling characteristic is intrinsic to the model, thereby avoiding inductive inequality constraints. The proposed method is characterized by its simplicity, high physical cohesion, good conditioning and computational efficiency. Keywords-Parameter estimation; RLCM model for transformer windings; frequency response analysis (FRA); multiobjective optimization; genetic algorithms; filament method; computational efficiencyI.

Detecting the location and severity of transformer winding deformation by a novel adaptive particle swarm optimization algorithm

“…In addition of improving the accuracy, phase function increases the convergence speed too. Even with all useful effects of the phase function, it is not often considered by researchers, for the sake of simplicity, increasing the convergence rate or evading a multiobjective problem . For simplifying the problem at first level, the number of considered frequency points decreased and only 1/10 of all frequency points are used at this level.…”

“…Different stochastic approaches based on measured sweep frequency response such as multiobjective GA, bee colony algorithm, bacterial swarming algorithm (BSA), improved particle swarm optimization, chaos optimization algorithm, genetic algorithm, particle swarm optimization, and bacterial foraging are used for solving the APDM problem. In most of the researches, for simplifying the solution of the problem and decreasing the size of the search space, a model with a low number of similar sections is considered and the number of used frequency points is largely reduced . In this paper, it is demonstrated that decreasing the number of used frequency points besides utilizing a model with a low number of sections leads to undesirable errors and an inaccurate model.…”

“…To compensate the weaknesses of these approaches, recently stochastic search methods are used for resolving the APDM problem. Different stochastic approaches based on measured sweep frequency response such as multiobjective GA, 19 bee colony algorithm, 20 bacterial swarming algorithm (BSA), 21 improved particle swarm optimization, 22 chaos optimization algorithm, 23 genetic algorithm, particle swarm optimization, and bacterial foraging 24 are used for solving the APDM problem. In most of the researches, for simplifying the solution of the problem and decreasing the size of the search space, a model with a low number of similar sections is considered and the number of used frequency points is largely reduced.…”

“…In most of the researches, for simplifying the solution of the problem and decreasing the size of the search space, a model with a low number of similar sections is considered and the number of used frequency points is largely reduced. 19,[21][22][23][25][26][27] In this paper, it is demonstrated that decreasing the number of used frequency points besides utilizing a model with a low number of sections leads to undesirable errors and an inaccurate model. Therefore, instead of those assumptions, using a fast, powerful, and efficient method is vital for extracting the model.…”

On the accuracy of detailed model inductance matrix estimation for air core winding

“…This method dose not calculate the inductance matrix. In [26] a multi-objective global search for solutions is performed by means of a genetic algorithm. An important point of parameters estimation in that paper is the selection of the initial values for parameters.…”