Modeling of a Signal Processing System for Aircraft Air Data Instrumentation

Speaker(s): 
Date: 
September 21, 1998 - 12:00pm
Location: 
90-3148
Seminar Host/Point of Contact: 

The velocity, attitude and altitude of flight vehicles have typically been measured with booms that extend from the vehicle surface out into the flow field.  However, this arrangement was found to be unacceptable for certain flight applications.  Instrumentation was therefore developed by other researchers to measure the flight parameters using an array of pressure measurements located on the surface of the vehicle.  The relationship between these pressure measurements and the air data is a complex non-linear function that is not easily described with simple aerodynamic models.  The focus of this work was to analyze this system, and develop modeling techniques to describe the functionality between measured and estimated parameters.  This system was characterized using both fluid dynamics and novel data processing techniques, and the system was subsequently modeled using a combination of physical and neural network models.  Statistical techniques were also developed to identify and isolate lost input signals.  The resulting systems of models were shown to be accurate and robust to noise in the input signals.  Additionally, the neural network models were shown to be stable across the entire relevant range of new input data, including both interpolation within and extrapolation outside the original domain of the training data.  Finally, back-propagation neural networks were shown to be well suited to this air data measurement technique.  The networks were shown to work well with the available vehicle geometry, and were found to be scalable to other vehicle shapes.

Add event to Google Calendar

Announcement List

To be added to our seminar annoucement list please, email ETA Seminars. Seminars may change without notice, stay informed by joining our seminar email list. 

Site Access

For site access please email your request to ETA Seminars. To arrive at Building 90 via the LBNL shuttle, take "Blue Uphill". Click here for the shuttle schedule. To ensure your site access is set-up send your request 36 hours in advance of the seminar.

Remote Access

Not able to attend in person? Remote participants can use ReadyTalk Audio & Net conference to phone in to listen, while viewing the presentation via the ReadyTalk site.

Web Access Code: 4864835

USA & Canada call-in number:
1 (866) 740-1260
Access Code: 4864835#

International callers:
Please visit this website to see if a toll-free call-in number is available for your international location.

If not, you will need to pay for the call yourself and use this phone number:

(303) 248-0285.