NASA and Boeing plan to begin flight tests of an advanced "neural network"-based controller on the F-15 ACTIVE (Advanced Control Technology for Integrated Vehicles) testbed by the end of next month. This is the first part of the Intelligent Flight Control System programme.
The neural network-based controller is expected to provide much more precise flight control than currently possible, and will be tested by NASA Ames in concert with Boeing's Phantom Works.
NASAACTIVE programme chief project engineer Gerald Schkolnik says: "The neural net system is an optimal controller. It will be used to develop optimal gains in real time. The neural network estimates all deviations and feeds the information to the flight control computer. This crunches on it with an on-line resolver, and responds accordingly."
The neural net tests are being hosted on the F-15 ACTIVE aircraft because "its computational capability makes it a ripe target for this sort of testing", says Schkolnik. The neural network will be hosted in a separate computer to the F-15's tri-channel VMS (vehicle management system) computers which host the research control laws. They are also segregated from the aircraft's quadruplex digital flight control computers that have been held in reversionary mode during tests of the ACTIVE's thrust vectoring system.
The first flight of the neural network base controller on the F-15 will be a NASA Level 1 milestone, says Schkolnik. After initial tests, the aircraft will revert to control by NASA Dryden, where it will continue with the ACTIVE programme. This is expected to include tests of the thrust vectoring system as a prime flight control system at speeds up to Mach 1.6.
Source: Flight International
