Sensor/Model Fusion for Adaptive Prognosis of Structural Corrosion Damage

Corrosion, stress-corrosion cracking or corrosion-initiated fatigue significantly impact maintenance downtime and structural life limitations of aging aircraft. Both legacy and new air platforms such as the Joint Strike Fighter (JSF), realize that corrosion will likely continue to be a structural challenge that warrants a structural health management system to provide accurate, cost effective assessments of a platform’s current (diagnosis) and future (prognosis) readiness. Corrosion/fatigue models exist that can reasonably predict failure progression in laboratory environments with controlled materials, usage profiles and environmental conditions. The prognostic challenge however, is to employ such models in the field where a priori factors and loading are far less certain and damage state awareness much more imprecise. With the goal of improving the accuracy of useful life estimates or time to inspection, an approach is presented in this paper for fusing imperfect state information such as global/local environmental measurements with physics of failure models to enable adaptive prognosis. Under the support of DARPA’s Structural Integrity Prognosis System (SIPS) program, a corrosion/fatigue growth model developed by Wei and Harlow of Lehigh University is adapted though calibration of initial conditions as well as internal state variables given measurements of temperature and periodic local damage estimates, using a technique known as Kalman filtering. When coupled with a stochastic wrapper, the prognostic model output provides time to a given structural damage level with confidence bounds from which informed operational and maintenance decisions can be made.