Prognostics of Power MOSFETs under Thermal Stress Accelerated Aging using Data-Driven and Model-Based Methodologies

Jose R. Celaya, Abhinav Saxena, Sankalita Saha, and Kai Goebel
Submission Type: 
Full Paper
Supporting Agencies (optional): 
phmc_11_009.pdf2.79 MBAugust 29, 2011 - 5:28pm

An approach for predicting remaining useful life of power MOSFETs (metal oxide field effect transistor) devices has been developed. Power MOSFETs are semiconductor switching devices that are instrumental in electronics equipment such as those used in operation and control of modern aircraft and spacecraft. The MOSFETs examined here were aged under thermal overstress in a controlled experiment and continuous performance degradation data were collected from the accelerated aging experiment. Die-attach degradation was determined to be the primary failure mode. The collected run-to-failure data were analyzed and it was revealed that ON-state resistance increased as die-attach degraded under high thermal stresses. Results from finite element simulation analysis support the observations from the experimental data. Data-driven and model based prognostics algorithms were investigated where ON-state resistance was used as the primary precursor of failure feature. A Gaussian process regression algorithm was explored as an example for a data-driven technique and an extended Kalman filter and a particle filter were used as examples for model-based techniques. Both methods were able to provide valid results. Prognostic performance metrics were employed to evaluate and compare the algorithms.

Publication Year: 
Publication Volume: 
Publication Control Number: 
Submission Topic Areas: 
Component-level PHM
Data-driven methods for fault detection, diagnosis, and prognosis
PHM for electronics
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed