Predictive reliability with signal based meta-models

Abstract

The development of algorithms and models to be used for prediction of the reliability and health monitoring of components and sensors is of great importance in aerospace, automotive and power generation industry. For this purpose metamodels have been developed that are based on physical simulations and that are able to quantify the impact of uncertainties on system behavior. These surrogate metamodels for time dependent signals can approximate the failure behavior and detect symptoms of aging. Furthermore, the prediction which input parameter combination can be run by the measurement setup without risk of failure or break in testing is an important application. Our approach has been validated for a high lift system in the aerospace industry.