Offline Parameter Estimation of a Fractional-Order Buck Converter Model

Abstract

Experimental studies have proven that real capacitors and inductors are better represented by fractional-order (FO) models. This is due to dielectric non-idealities, Eddy current losses, and hysteresis effects. This fact has motivated many studies on the modeling and analysis of FO converters, ie., converters with fractional capacitors and inductors. On the other hand, parameter identification of power electronic systems is an area of increasing importance. This is due to its application to adaptive control and condition monitoring. This paper investigates the problem of parameter identification of FO buck converter in continuous conduction mode (CCM). To this end, the Trigeassou approximation of the FO integral is compared with the predict-evaluate-correct-evaluate (PECE) method to justify its use in this model in terms of accuracy and computation time. The identification problem is formulated based on non-invasive measurement quantities (input current and output voltage). Four cases of synthetic data, with added noise, are used to validate the effectiveness of the proposed identification procedure. The cuckoo search optimizer (CSO) is used to identify the parameters, and it demonstrates outstanding consistency and accuracy in the results across independent runs.