Air data sensing from surface pressure measurements using a neural network method

Neural networks have been successfully developed to estimate freestream static and dynamic pressures from an array of pressure measurements taken from ports located flush on the nose of an aircraft. Specific techniques were developed for extracting a proper set of neural network training patterns from an abundant archive of data. Additionally, the specific techniques used to train the neural networks for this project were reported, including the scheduled adjustments to learning rate parameters during the training process. The performance of the trained networks was compared to the accuracy of the aerodynamic model that is currently being applied to these flush air data sensing systems.