Summary

Description

A prognostic system needs to separate nominal component behavior from the faulty ones even in the cases when those behaviors are similar. Advanced pattern recognition techniques are required to separate nominal and faulty input-output component data vectors in a complex high-dimensional space. We propose to develop the Multi-agent Prognostic Health and Usage Monitoring (Multi-PHUM) and test it in a subsection of an aerospace vehicle. MULTI-PHUM is hierarchical with the lower levels performing ordinary diagnostics and prognostics using graph-based fault diagnosis technique to place alarms on safety-critical components and handle situations with multiple faults. At the intermediate level of hierarchy, MULTI-PHUM uses neural network techniques such as the Extended Auto Associative Neural Networks (E-AANN) to detect the faults not detected by lower level graph-based method. Yet at a higher level of hierarchy, MULTI-PHUM performs advanced rule-based pattern recognition for abnormalities in the system not detected by lower levels. The essence of MULTI-PHUM is based on the latest tools in the information age and hence has a strong commercial potential for the management of other systems that have to economically maintain healthy fleets such as airline systems with many aircrafts or future swarms of UAVs.