Page 214 - Neural Network Modeling and Identification of Dynamical Systems

6.2 SEMIEMPIRICAL MODELING OF LONGITUDINAL SHORT-PERIOD MOTION FOR A MANEUVERABLE AIRCRAFT 205

FIGURE 6.5 Semiempirical (gray box) ANN model of the longitudinal angular motion of the aircraft (based on Euler
difference scheme).

TABLE 6.1 Simulation error on the training set (polyharmonic signal).
Problem Point mode Monotonous mode
RMSE α RMSE q RMSE α RMSE q
Adjusting C L 1.02 · 10 −3 1.24 · 10 −4 1.02 · 10 −3 1.24 · 10 −4
Learning C L 1.02 · 10 −3 1.23 · 10 −4 1.02 · 10 −3 1.24 · 10 −4
Learning C L , C m 1.02 · 10 −3 1.19 · 10 −4 1.02 · 10 −3 1.27 · 10 −4
NARX simulation 1.85 · 10 −3 3.12 · 10 −3 1.12 · 10 −3 7.36 · 10 −4

TABLE 6.2 Simulation error on the test set (polyharmonic signal).
Problem Point mode Monotonous mode
RMSE α RMSE q RMSE α RMSE q
Adjusting C L 1.02 · 10 −3 1.59 · 10 −4 1.02 · 10 −3 1.17 · 10 −4
Learning C L 1.02 · 10 −3 1.59 · 10 −4 1.02 · 10 −3 1.17 · 10 −4
Learning C L , C m 1.02 · 10 −3 1.32 · 10 −4 1.02 · 10 −3 1.59 · 10 −4
NARX simulation 2.32 · 10 −2 4.79 · 10 −2 3.16 · 10 −2 5.14 · 10 −2

that the curves on the graphs practically coin- timate of the model accuracy on the training
cide. Quantitative estimates of the accuracy for set) and Table 6.2 (estimate of the generalization
the obtained models are given in Table 6.1 (es- properties for the ANN model), along with the