Search

With the increase in power electronic equipment in power system, the excitation of ferromagnetic materials often involves a large number of harmonics. Therefore, it is necessary to construct an accurate dynamic hysteresis model to adapt to this complicated operating state of electrical equipment. In this paper, a Hybrid Dynamic Hysteresis Model (HDHM), which can effectively characterize the harmonic excitation of materials is studied based on the Preisach model and Stacked Auto-Encoder (SAE) model. The static part of this model takes the form of the inverse Preisach model. And the Multiple Dynamic Hysteresis Model Set (MDHMS) is constructed by multiple dynamic models of eddy currents and excess characteristics of the ferromagnetic materials. The dynamic part of the HDHM takes the form of the model structure combining the Stacked Auto-encoder and the MDHMS. The calculation results of the hysteresis loop and ferromagnetic loss in the harmonic condition of silicon steel sheet proves the validity of this model. Moreover, compared with the conventional dynamic hysteresis model, the HDHM has better accuracy and generalization ability. [ABSTRACT FROM AUTHOR]

• The HIHM has the better structure and higher precision than traditional models. • The magnetization state vector is used to characterize the specific magnetization of materials. • The modeling of HIHM can reduce the requirement of hysteresis data. • The HIHM can overcome this unreasonableness in the modeling of Preisach. • To the parameter identification of the J-A model, the training of HIHM taken less time. The hysteresis modeling of ferromagnetic materials in electrical equipment is one of the basic theoretical studies in the field of electrical engineering. Long training time and poor accuracy are the main challenges of the traditional hysteresis models. In this paper, a Hybrid Intelligent Hysteresis Model (HIHM) is proposed by combining the hysteresis operator space theory and Deep Belief Networks-Deep Neural Network (DBN-DNN) algorithm. The Preisach Fusion Operator (PFO) is constructed by fusing multiple traditional Preisach operators with coefficients satisfying the Gaussian distribution. And multiple PFOs are used to construct an operator space to create the high-dimensional operator data. By the operator space, the nonlinear hysteresis relationship in the hysteresis data is transformed into a nonlinear mapping between the operator data and the output of the model. The DBN-DNN model is used to characterize this nonlinear mapping. The structure of the HIHM is determined by taking the output of the operator space as the input of the DBN model. The generation of first-order reversal curve proves that the model has certain practicality. In addition, the validity and generalization of this model is verified through the simulation with experimental data of non-orientation (NO) silicon steel B35A210 at quasi-static frequency. [ABSTRACT FROM AUTHOR]

• Based on the white-box model theory, the Stacked Auto-Encoder is used to identify the relevant parameters of this model. • The anisotropy compensation coefficient is introduced into the model structure to reduce the model error. • The phase angle compensation parameters are calculated by the magnetic loss data to improve the loss calculation accuracy. • The Neural Networks construct the mapping between model structure and the output of hysteresis data. • The median iterative algorithm is used to search the corresponding compensation angle values. The construction of the vector hysteresis model is significant for the electromagnetic characteristics analysis of electrical equipment. With a deep learning parameter identification algorithm, this paper presents a modeling method of the inverse vector hysteresis model. The vector Everett function is obtained by the alternating hysteresis data. In order to reduce the error of the vector Everett function, the anisotropy compensation coefficient is introduced into the model structure. Based on the white-box model theory, the Stacked Auto-Encoder (SAE) model is used to identify the relevant parameters of the inverse vector hysteresis model. And the parameters of the phase angle compensation function are calculated by the magnetic loss data to further improve the magnetic loss calculation accuracy of the vector model. Compared with the experimental data of rotating magnetization, the fitting result proves the accuracy of the model in a wide range. In addition, the validity of the model is verified by calculating the dynamic magnetic loss. [ABSTRACT FROM AUTHOR]