Application of neural network technology to fly-by-light control systems

摘要

Abstract: McDonnell Douglas Aerospace (MDA) is developing aNeural Network based Intelligent Flight Control Systemthat utilizes fly-by-light data transmission combinedwith high capacity flight processors to implementcontrol advancements and fault monitoring processes. Inthis control system design, high speed data transferand electromagnetic interference protection obtainedthrough fiber optic technology is linked with NeuralNetwork based flight control hardware processors thatare programmed with damage adaptive control capability,thus providing maximum survivability for fighteraircraft. This system also provides enhanced componentfault diagnostics that can identify subsystem failuresduring flight, thus providing reduced life cycle costthrough efficient maintenance action and less downtimeof the aircraft. The Intelligent Flight Controlproducts apply to fighter and transport aircraft. Thisprogram is Task 2C of the ARPA Fly-by-Light AdvancedSystems Hardware (FLASH) Technology ReinvestmentProgram. The principal partner with MDA for the Task 2CIntelligent Flight Control development is MartinMarietta Control Systems. The program will mature thesystem hardware and software for laboratorydemonstrations of component fault diagnostics andhighly adaptive flight control performance.!0