NEURAL NETWORK TECHNOLOGY FOR JET ENGINES

Integrated flight and propulsion control systems will be necessary to satisfy critical mission requirements in future aircraft. Aircraft flight and propulsion control systems are traditionally designed to operate independently of each other. This method often results in a system where performance is compromised for robustness and operability. There exists a great need for nonlinear system identification and control approaches that can optimize on-board models for off-nominal behavior such as engine deterioration and engine-to-engine variations. NeuroDyne will team with Scientific Systems in the development of a nonlinear adaptive engine diagnostics and control method using a combination of neural network based system identification and optimal control for aircraft engines. This program will leverage an existing program funded by NASA Lewis and Dryden Research Centers which involves NeuroDyne and McDonnell Douglas in the development of a neural network based estimation component for engine efficiencies and diagnostics in the Performance Seeking Control (PSC) system. Our proposal will extend these results to develop a nonlinear system identification and optimal control component for the Pratt and Whitney 1128 engine operating in a F-15 aircraft.