Closed-Loop Control Learning - Neural Networks: Methodology and Applications - page 298

Fig. 5.4. Trajectory of the stabilized system with a noise factor of 3. (Straightfor-
ward model inversion control learning with larger weighting of the angle deviation
in the cost function)
with the desired time constant. On the previous example of inverted pendu-
lum, that method gives much better results in the neighborhood of equilibrium
with the same model. A typical trajectory of the closed-loop controlled system
is shown on Fig. 5.6 with a control noise factor equal to 3.
Another neural controller design method is proposed in [Levin 1993]. That
method is somewhat similar to the previous method; it consists in choos-
Fig. 5.5. Learning process of a reference model controller