基于改进Preisach模型的软磁复合材料复杂励磁损耗计算方法OA

Soft magnetic composites(SMCs)have shown significant potential for applications in key magnetic components of power electronic converters(e.g.,DC inductors,high-frequency transformers)owing to their distributed air-gap structure and high saturation density.However,the loss characteristics of such materials under complex excitation states(DC bias,non-sinusoidal excitation)are difficult to accurately assess using traditional methods,thereby restricting device design and optimization.This paper proposes an improved Preisach hysteresis model and an eddy-current loss decomposition method for calculating hysteresis and eddy-current losses in soft-magnetic composites. In hysteresis loss calculation,the traditional limiting-hysteresis-loop Preisach model requires only a single saturated hysteresis loop.Still,its prediction accuracy is limited in the low-density region.This paper proposes a multiple-hysteresis-loop correction method to refine the distribution function of the hysteresis operator in the Preisach plane by incorporating multiple sets of measured hysteresis-loop data,including the low-density region.Specifically,the Preisach plane is discretized into grid cells,and the grid's magnetic moment density distribution matrix is adjusted using measured data to improve the model's accuracy for hysteresis loops at low magnetic densities and under DC bias.The results show that the improved model's error in the range 0.3～1.0 T is lower than that of the original model.It exhibits more stable computational performance under DC bias conditions,outperforming the traditional Jiles-Atherton model in predicting coercivity and remanent magnetization. For eddy-current loss,this paper decomposes the total eddy-current loss into two components,intra-particle and inter-particle losses,based on the micro structural characteristics of soft-magnetic composites.The local eddy current effect of the magnetic powder in the ferromagnetic phase causes intra-particle loss.The loss density is modeled as a function of the geometric parameter and the magnetic powder's conductivity.At the same time,macroscopic conductive paths generate interparticle losses due to incomplete cladding of the insulating layer,which are quantified by the material's homogenized conductivity and shape factor.The eddy-current loss characteristic parameter ξ is proposed.It depends solely on material properties and simplifies loss calculations under non-sinusoidal excitation,such as square-wave or triangular-wave excitation. The energy storage inductor in a Boost converter and the high-frequency transformer in an LLC resonant converter are used as platforms,both designed for a switching frequency of 30 kHz.Pure iron powder cores are prepared by the sol-gel method,and the inductor losses in continuous and discontinuous conduction modes,as well as the core losses of the high-frequency transformer under the same magnetic density ripple,are predicted by combining the multiple-hysteresis-loop Preisach model with the eddy-current loss formula.The experimental results show that the proposed improved method achieves a total loss prediction accuracy of no more than 5%relative to the traditional method in complex excitation scenarios,such as single-ended excitation.In addition,the computational efficiency is high,with a single hysteresis-loss solution taking approximately 8.7 s.It can be integrated into the FEA post-processing workflow and is suitable for loss assessment of complex-shaped cores.