A complete hardware package for a fault tolerant flight control system using online learning neural networks

Abstract

This paper shows the results of a research effort focused on demonstrating the capabilities of hardware based online learning parallel neural networks for a fault-tolerant flight control system. Particularly, for a given aircraft mathematical model, two different fault-tolerant schemes have been implemented in different neural networks embedded on a mother-board with 4 TMS320C40 DSPs. The first scheme provides sensor failure detection, identification, and accommodation (SFDIA) for different types of sensor failures within a flight control system assumed to be without physical redundancy in the sensory capabilities. The second scheme provides actuator failure detection, identification and accommodation (AFDIA) for different actuator failures. Emphasis has been placed to ensure real-time capabilities as well as an efficient integration between the AFDIA and the SFDIA schemes without degradation of performance in terms of false alarm rates and incorrect failure identification. The results of the simulation following different types of failures are reported.