Development and Evaluation of a Multi-Agent System for Gas-Turbine Engine Health Management

The development of Unmanned Aerial Vehicles (UAVs) is continuously pushing the boundaries for more autonomous systems. In order to make realistic the goal of autonomous civil UAVs flying within civil airspace, technology must be developed so that UAVs and all their critical subsystems can be certified for autonomous flight. In this paper, we address specifically the Propulsion subsystem and the problems that arise when engines are affected by failures. In a Pilot-Out-Of-theLoop situation, the engine controller must be able to autonomously detect, isolate and eventually mitigate failures, which involves making decisions that are delegated to the pilot in current systems. We propose a Propulsion Health Management System for a general twin-engine UAV, based on Intelligent Agent technology. The system fuses different software techniques to achieve the Fault Evaluation and Fault Mitigation functions, which would normally be performed by the pilot after Fault Detection and Fault Isolation. The proposed system is described in detail and then the results of simulation tests are presented.