APPROXIMATION OF FREQUENCY DOUBLER MAGNETIZATION CURVES BASED ON A SECOND-ORDER POLYNOMIAL

Abstract

The article is dedicated to the relevant problem of increasing the efficiency of ferromagnetic frequency doublers, which are important elements of modern high-frequency electronic systems. The relevance of the study is due to the growing demands in the modern world for performance, energy efficiency, and miniaturization of electronic devices. The aim of this work is to maximize the load current of a ferromagnetic frequency doubler. The optimization algorithm chosen is the genetic algorithm, which performs well in solving nonlinear problems and has the ability to find a global optimum in a complex multidimensional parameter space. A mathematical model of a three-rod ferromagnetic frequency doubler was used for the study. In this paper, a polynomial approach based on a second-order polynomial is proposed to approximate the magnetization curves, instead of the classical approach of approximation using a cubic spline. The cubic spline approximation has three sections: linear, nonlinear, and linear sections. The polynomial approach, in turn, contains only two sections: linear and nonlinear sections. In this approach, the coefficients of the polynomial are calculated relative to the end of the linear section, which is due to the need for the nonlinear section to continue where the linear section ends. One coefficient of the polynomial must be defined, for example, by a genetic algorithm, and the other two are calculated relative to the first. In this paper, the optimization was performed in three rounds. In the first round, the basic values of the frequency doubler were used, and the genetic algorithm searched for the polynomial coefficient. The second round additionally searched for input voltages. In the third round, in addition to the previous one, the search area was expanded to include the inverse inductances of the magnetic windings. The results obtained indicate a 15% increase in load current using the polynomial-based approximation compared to the cubic spline-based approximation.