Embedded Reasoning Supporting Aerospace IVHM
Abstract
The work presented in this paper summarizes an effort to advance the current state of vehicle health reasoning practice as a fundamental element of an Integrated Vehicle Health Management (IVHM) approach. This effort seeks to integrate detection, diagnostic, and prognostic capabilities with a hierarchical diagnostic reasoning architecture into a single synergistic, embeddable commercial off-the-shelf (COTS) unit. Three avenues have been pursued to advance the capabilities of health reasoning in support of IVHM. First, the scope of the systems/subsystems being monitored by the underlying Health and Usage Monitoring System (HUMS) has been expanded. The HUMS is comprised of subsystem specific prognostic and health management (PHM) modules developed to process raw measured data and supply condition indicators (CI) targeting several flight critical areas from advanced aerospace applications: such as electro-mechanical actuators (EMA), engine performance, power train vibration, and structural impact detection and isolation. Second, a hierarchical, model-based reasoning approach has been developed to isolate and classify latent failure mode manifestations existing in the condition indicators output by the diagnostic and prognostic algorithms and utilize this information to assess the functional availability of the vehicle and its constituent subsystems. At the lowest level, the embedded reasoning seeks to classify latent failure mode indications from raw sensor data or condition indicators considered as evidence sources and isolate the root cause mechanism along with the accompanying severity. In addition, a confidence assessment is determined by considering the temporal element and how many times a particular piece of indicting evidence has repeated in the previous N evaluation iterations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2007
- Accession Number
- ADA498088
Entities
People
- Mark Derriso
- Matthew S. Leonard
- Michael H. Schoeller
- Michael J. Roemer