Neural Identification of Controlled Dynamical Systems and Recurrent Networks - Neural Networks: Methodology and Applications - page 271

Fig. 4.14. Recurrent input-output neural network for an open loop controlled dy-
namical system modeling
dynamical system, a controller may be modeled as an application of the state
space onto the control space, which associates the appropriate control to the
current state. The diagram of Fig. 4.15 represents the connection of these two
neural networks.
In that diagram, the state input feeds both the process model and the
controller, which computes the control signal. That control signal is the second
Fig. 4.15. Recurrent neural network modeling a closed loop controlled dynamical
system. The model is the combination of network ϕ RN and network ψ RN