Early diagnosis in DC-link capacitors: electrolytic and films

This chapter presents an RUL prediction algorithm based on accelerated life test data and derived physics models for electrolytic capacitors. The main elements are (1) development of the first principles based degradation models; (2) the implementation of a Bayesian-based health state tracking and RUL prediction algorithm based on the Kalman filtering framework. One major advancement reported here is the prediction of RUL for capacitors as new measurements become available. The key contribution of this work is the prediction of RUL for capacitors as new measurements become available. The derived degradation models can be updated and developed at a finer granularity to be implemented for detailed prognostic implementation. This capability increases the technology readiness level ofprognostics applied to electrolytic capacitors.The results presented here are based on accelerated life test data and on the accelerated life timescale. Further research will focus on development of functional mappings that will translate the accelerated life timescale into real usage conditions timescale, where the degradation process dynamics will be slower and subject to several types of stresses. The performance of the proposed exponential-based degradation model is satisfactorily based on the quality of the model fit to the experimental data and the RUL prediction performance as compared to ground truth.