Multi-step Least Squares Algorithm for Thermal Characterization Based on Mission Profile

The ageing of power modules (PMs) used in power electronics applications changes the electrical and thermal parameters over their lifetime; thereby increasing the power losses and thermal impedance, and affecting the thermal model (TM) accuracy. Thus, PM ageing is accelerated and reliability degraded due to the overheating of semiconductor components. This paper deals with mission profile based analysis to identify a PM thermal digital twin (TDT). Since TM contains both slow and fast time constants, the conventional ordinary least squares (LS) minimizing one-step error prediction has poor performance in identifying of TDT parameters. The proposed algorithm is an extension of LS. The ability of the proposed algorithm to find optimal parameters is enhanced by adding additional terms penalising the quadratic error of multi-step prediction. The method allows TM parameters to be track over the PM lifetime, enabling condition monitoring and accurate die temperature estimation. The algorithm is able to identify both slow and fast time constants with accuracy of 1.5 °C.