Developing Damage Prognosis Solutions

摘要

Damage prognosis is the prediction in near-real-time of the remaining useful life of an engineered system given the measurement and assessment of its current damaged (or aged) state and accompanying predicted performance in anticipated future loading environments. Los Alamos National Laboratory (LANL) along with research partners at Stanford University and Virginia Tech has launched a large-scale, multi-disciplinary program that to develop and integrate advanced microelectronic sensing technology, tera-scale predictive modeling capabilities for damage evolution, and techniques for correlation-based data compression to address the problem of damage prognosis in aerospace, civil, and mechanical engineering systems. These Damage Prognosis solutions realize information fusion between a robust, densely populated microelectronic sensing array and a system-specific predictive modeling capability derived from benchmarked high-fidelity tera-scale models. The tera-scale predictive models will be compressed to obtain a near one-to-one correspondence with the measurement degrees of freedom. Such compression will allow the surrogate predictive models to be deployed on the monitored system using advanced micro-electronic processing hardware that is directly coupled with the sensing array. This paper will begin by discussing the hardware and software components of this system and the current proof of principal experiments being conducted at Los Alamos. The paper will also provide a brief summary of the Accelerated Strategic Computing Initiative that has built the world's most powerful computers and will show how this computer is being utilized for damage prognosis.