System Identification via Artificial Neural Networks: Applications to On-line Aircraft Parameter Estimation * 975612

In this report, the neural identification problem is outlined and the identifiability question for a general class of recurrent neural networks is addressed. As part of the intelligent flight control concept program, recurrent second-order neural networks are utilized in order to continuously identify critical stability and control parameters during flight. Our group at Washington University participated in Phase II, the online learning, with neural networks that learn new information during flight. In particular, a recurrent second-order neural network architecture with a robust filtered error learning algorithm was utilized to identify the dynamics of an F-15 aircraft.
While the emphasis of our work has been on the development and implementation of online neural network estimators, we shall also include results with and without the baseline network. Several examples including in-flight situations are presented and the effectiveness of the recurrent high-order neural networks is illustrated.


Subscribers can view annotate, and download all of SAE's content. Learn More »


Members save up to 43% off list price.
Login to see discount.
Special Offer: With TechSelect, you decide what SAE Technical Papers you need, when you need them, and how much you want to pay.