A Wavelet descriptor model of hysteresis loop phenomena. Issue 11 (18th August 2020)

• Record Type: Journal Article
• Authors: Bisht, Hrishitosh; Gadre, Vikram M.; Kulkarni, Shrikrishna V.; Sai Ram, Boggavarapu
• Is Part Of: IET electric power applications. Volume 14:Issue 11(2020)
• Page Start: 2179
• Page End: 2186
• Publication Date: 2020-08-18
• Subjects: finite element analysis -- magnetic hysteresis -- Fourier transforms; wavelet descriptor model -- hysteresis loop phenomena -- transient finite element analysis -- Fourier descriptor method -- magnetic hysteresis curves -- wavelet descriptors -- hysteresis losses -- wavelet basis -- Fourier basis -- B/H curve -- higher‐order harmonics -- pulse‐width modulated source; Electric power -- Periodicals; Electric power systems -- Periodicals; 621.305
• DOI: 10.1049/iet-epa.2020.0363
• Languages: English
• ISSNs: 1751-8660

Abstract

Abstract : An accurate and computationally efficient hysteresis model is required to perform transient finite element analysis of electrical machines and transformers. Numerical experiments performed in the current work suggest that there are severe limitations of the Fourier descriptor method. In this work, the authors build upon the idea of Fourier descriptors and propose a wavelet descriptor model to characterise magnetic hysteresis curves, with the advantages of Fourier descriptors while their disadvantages are mitigated. It is shown that wavelet descriptors can also be used to compute hysteresis losses. The motivation behind using a wavelet basis rather than a Fourier basis is the ability of wavelets to represent local features of the signal. The method can efficiently represent a complete B/H curve with only a few data‐points and allows for an automated representation of hysteresis loops. Wavelet descriptors are used for the first time to model a wide variety of hysteresis loops with asymmetric minor loops generated using higher‐order harmonics, and from a pulse‐width modulated source. The results obtained using the proposed method are compared with those of the Fourier descriptor method. In all cases, the proposed method outperforms the Fourier descriptor method. The proposed method is computationally efficient and simple to understand.