Reliability prediction of power electronic converters based on surrogate models

Power electronic converters (PEC) are integral to various sectors, including power systems, transportation, and industrial production, where their reliability significantly influences overall system performance. Despite its importance, current research has not thoroughly addressed the simultaneous impact of component performance degradation and functional failures on electronic system performance. This paper introduces a reliability prediction method for PEC that considers both component performance degradation and functional failure. Initially, digital prototype models of PEC are developed to accurately represent the operational processes and simulate performance under each critical fault mode. To minimize simulation costs, two surrogate models based on neural networks are then established for PEC. Subsequently, through iterative calculations that utilize models of performance degradation and functional failures of components, an analysis of both PEC performance degradation and functional failure states is conducted, facilitating the completion of PEC reliability prediction. This proposed framework for reliability prediction may also be applicable to other electronic systems.