Figures & data
Figure 1. Strain gauges installed on the aft fuselage in the vicinity of the vertical tail.
Figure 2. Locations of strain gauges used to train the neural network.
Figure 3. Strain gauge data used as input data to train the network.
Figure 4. Magnification of strain gauge data illustrating the effect of buffet loading.
Figure 5. Strain data before and after filtering.
Figure 6. Model of relationship between aerodynamic loading and strain response.
Figure 7. Model of inverse relationship between strain response data and aerodynamic loading.
Table 1. Common network types
Table 2. Summary of results for several networks
Figure 8. Layout of the selected Elman neural network.
Figure 9. The layout of the first and second layers in the selected Elman network.
Figure 10. Diagram of the output layer in the selected Elman network.
Figure 11. Diagram of the complete Elman network.
Figure 12. Estimated load compared with actual load for sample training data.
Figure 13. A comparison of estimated manoeuvre load to actual manoeuvre load.
Figure 14. A comparison of estimated buffet load to actual buffet load for the training data.
Figure 15. A comparison of estimated buffet load to actual buffet load for the test data.
Figure 16. Locations of load actuators, strain gauges and ‘unknown load’.
Figure 17. Estimated load at ‘new location’ compared to known actual load at that location.
Figure 18. Plot of instantaneous loading across vertical tail following interpolation of known loads.
Figure 19. Estimated load at ‘new location’ compared to known load in region of training data.
Figure 20. Estimated load at ‘new location’ compared to known load for test data.
