Yangkai Che studies electromagnetic nondestructive testing and physics-informed neural networks. The work addresses the ill-posed inverse problem of reconstructing a magnetic field from sparse observations under bias and disturbance. It combines electromagnetic constraints with measured data to localize metal defects and recover their shape for inspection in complex working conditions.
主要研究方向为电磁无损检测与物理信息神经网络，针对偏置扰动下稀疏磁场观测带来的逆问题病态性，开展机理-数据协同的磁场鲁棒重构与金属缺陷定量反演研究。以电磁场物理约束为基础，构建物理相容的反演框架，提升缺陷定位与形貌重建精度，为复杂工况下智能电磁检测提供方法支撑。