On the Integration of Real-Time Diagnosis and Prognosis for Scheduled Maintenance Optimization

Abstract

Helicopters are very critical aircrafts from the point of view of fatigue loads. A structural damage could grow fast to critical size because of the wide range of manoeuvre loads. As a consequence, scheduled maintenance is a major cost during the operative life of the aircraft. The most advanced methodologies for Structural Health Monitoring and Prognostic Health Management available today are aimed to provide a real-time structural diagnosis, thus maximizing the availability of the helicopter and reducing the maintenance costs. However, on-board diagnostic systems might be gradually introduced in the current maintenance procedures, trying to minimize the risk associated to these newly developed technologies. The work presented inside this paper is about the simulation of the integration of real-time diagnosis and prognosis into a typical scheduled maintenance procedure. A diagnostic unit is capable for anomaly detection and damage quantification (in terms of crack length). It is trained with Finite Element simulated damages and tested with real experimental data. The diagnostic output is then processed in a particle filter algorithm, based on sequential importance sampling technique, aimed at refining the estimation of the structural condition as well as to update the inference on the residual useful life distribution. The coupling of real-time diagnosis with off-line measures (taken during scheduled maintenance stops) is analyzed and applied to a damage tolerant structure, trying to outline the advantages and drawbacks of the proposed approach.