Adaptive Identification and Thyristor Converter Control

Abstract

The thyristor converter (TC) current control problem with use made of an ‘input-output’ model when affected by random disturbances in a measuring system is considered. It is shown that in the simplest case the current model may be presented in the form of a linear combination of exponential and harmonic terms with a priori unknown coefficients. It is noted that the obtained model is nonlinear relative to unknown parameters. To linearize it use is made of the differential transformation of current measurements which allows a linear combination of exponential terms to be selected from the initial model. It is emphasized that there is instability of calculations when identifying unknown parameters of the exponential component of the model. To improve stability there is suggested a method of natural regularization by selecting the current sensor sampling frequency and the number of measurements. Exponential smoothing of estimates of the electromagnetic time constant in a load circuit and the algorithmic filtration of current measurements with the use of spline functions are also advantageous.