As a core component of power conversion systems

the reliability of power modules directly determines the safety and stability of systems such as renewable energy generation and high-voltage direct current (HVDC) transmission. Multi-physics modeling and analysis are essential for revealing the operating states and degradation mechanisms of power modules. However

traditional numerical methods face computational efficiency bottlenecks

making it challenging to meet the real-time condition monitoring and intelligent diagnostic requirements of modern power systems. Therefore

this paper summarizes the packaging forms and multi-physics modeling methods of bond-wire and press-pack power modules. Tthe research progress in fast solution techniques for power modules is reviewed

focusing on model order reduction

projection-based reduction

and deep learning-based reduction methods. Moreover

this paper also discusses the local reduction methods for multi-physics coupled models and provides a comparative analysis of the advantages and disadvantages of various fast solution methods. Based on these discussions

this paper further explores the urgent research areas

offering significant reference value for advancing multi-physics fast solution methods

digital twins

and smart operation and maintenance.