Optimal fitting of high-frequency cable model parameters by applying evolutionary algorithms

Due to the widespread use of electronic power converters, low-voltage high-frequency cable models are being increasingly applied in industry, automobile or aeronautics applications among others. It is known that depending on switching frequency, cable configuration and length, transient overvoltage effects comprising a wide frequency range from dc up to several tens of MHz can appear. However, to accurately reproduce the wide-band frequency response, such models often require the use of ladder networks, thus being necessary to adjust the values of a relatively large number of R, L and C components, which is a complex task. This paper is focused to solve this problem, which is done by applying an iterative genetic algorithm (IGA) optimization approach. From a set of experimental short circuit and open circuit tests the high-frequency cable model of a given cable configuration is obtained, whose parameters are fitted by means of the proposed IGA-based method. Finally, the accuracy of the model obtained is validated experimentally by comparing the frequency-domain and time-domain responses through overvoltage predictions of different samples of the analyzed cable.