Page 147 - Neural Network Modeling and Identification of Dynamical Systems
P. 147
4.2 PERFORMANCE EVALUATION FOR MULTILAYER NEURAL NETWORK MODELS OF AIRCRAFT MOTION 137
FIGURE 4.4 Neural network model design for micro-UAV “003,” flight regime with airspeed V = 30 km/h. (A) Training
sample for the ANN model. (B) Efficiency test for a closed-loop ANN model. Here α is the angle of attack, deg; E α is error
(difference between angle of attack values for the object and the ANN model), deg; δ e, act is command signal for the actuator
of the elevons; t is time, sec; the solid line is the output of the object; the dotted line is the output of the ANN model.