Real-time temperature sensitive electrical parameters and model- based condition monitoring for PV inverter applications

Abstract

To improve the reliability of power converters, condition monitoring is a useful approach as this allows to schedule maintenance more carefully or even implement adaptive control. The current state of the art is focused on the measurement of damage sensitive parameters which can be used directly or complementary with a digital twin. Several papers report successful experiments where these damage sensitive parameters of switching devices are used to evaluate wear under constant power conditions. This approach needs to be extended for PV inverter applications as these experience complex power profiles in varying weather conditions. This poster presentation will focus on a methodology to monitor the damage sensitive drain to source on-resistance in concession with the internal gate resistance which is only sensitive to the junction temperature. This allows to separate the influence of temperature and damage which allows the use of complex mission profiles. To simplify the evaluation even further, a second approach is proposed where the additional gate resistance measurement is replaced by a modeling strategy. The electrical part of the electro-thermal model is build using a two dimensional lookup table approach and the thermal model is represented by a Cauer network. The thermal resistances in each layer are determined by a FEM simulation. This poster presents the preliminary results which show a validation of the model with measurements on an inverter prototype and the proposed condition monitoring circuits simulated in spice.