How MagNet: Machine Learning Framework for Modeling Power Magnetic Material Characteristics

Abstract

This paper applies machine learning to power magnetics modeling. We first introduce an open-source database – MagNet – which hosts a large amount of experimentally measured excitation data for many materials across a variety of operating conditions, consisting of more than 500,000 data points in its current state. The processes for data acquisition and data quality control are explained. We then demonstrate a few neural network-based power magnetics modeling tools for modeling the core losses and B–H loops. Machine learning allows multiple factors that may influence the magnetic characteristics to be modeled in a unified framework, while provides insights to quantify the complexity of magnetic characteristics and reduce the size of the measurement data required to build a precise model. Neural network models are found to be effective in compressing the measurement data and predicting the material characteristics. The behaviors of a typical power magnetic material (TDK N87) across a wide range of operating conditions (e.g., temperature, waveform, dc-bias) can be well described by a small-scale neural network (204 KB) which is 2,500 times smaller than the raw measured time-series data (512 MB), paving the way for “neural networks as datasheet” to assist power magnetics design.