SiC MOSFETs Junction Temperature Estimation Method Considering Aging Influence

Junction temperature (T
j
) is a critical parameter for condition monitoring and lifetime assessment of SiC MOSFETs. As one of the temperature-sensitive electrical parameters (TSEPs), the on-state drain-source resistance (R
on
) can be calculated by measuring V
ds
and I
d
, making it a widely used T
j
indicator in SiC MOSFETs. However, the relationship between R
on
and T
j
may change with aging of SiC MOSFETs, leading to errors in T
j
estimation. Therefore, an improved SiC MOSFETs T
j
estimation method considering the influence of aging is proposed to eliminate this error. Back Propagation Neural Network (BPNN) is employed to map the work accumulation to an aging factor, so that to correct the result of R
on
-based T
j
estimation method. To validate proposed method, a dc power cycling test platform is built. The number of power cycles is taken as work accumulation, which represents the aging state of SiC MOSFETs. The test results shows that the R
on
-based T
j
estimation method without correction of can result in errors of up to 18°C after 6337 power cycles. However, the proposed method is able to reduce the errors to less than 2°C at the same aging state, confirming the effectiveness of the proposed method.