Mission Profile Clustering Using a Universal Quantile Criterion

The current trend in the automotive industry towards increasingly detailed and more granular mission profiles (MPs) is also giving rise to an enormous number of various MPs, distinguishing each individual application and use case imaginable. To keep an overview in the face of this flood of data, to filter out and select the qualification relevant MPs, and to create a suitable and simply designed reliability requirement, a MP clustering approach using a quantile criterion is presented. This generic approach is universally applicable and therefore not limited to the automotive industry.Based on equivalent test times (ETTs) and acceleration factors (AFs), after determining the quantile criterion, a representative MP can be selected either from an existing best fit MP or an artificially generated MP, obtained by numerical methods. Furthermore, the coverage of these MP requirements by technology reliability capabilities can be evaluated with ease. This approach is a generic concept and can therefore be applied to all well-known and commonly used damage accumulation and failure acceleration models. In the presented automotive case studies, the temperature accelerated Arrhenius model, multidimensional MPs with temperature and voltage acceleration, the temperature cycling Coffin-Manson model as well as a non-linear damage accumulation model are investigated.