Design of neural network PID controller based on E-FRIT

When a controlled object has a nonlinear characteristic, a good control result is not always obtained with fixed PID gains. This paper proposes a design method of a nonlinear PID controller using a neural network to overcome the problem. In the proposed controller, PID gains are tuned online by a neural network and a controlled object is controlled by the PID controller with the tuned PID gains. The neural network learns by a learning algorithm based on the Extended-Fictitious Reference Iterative Tuning (E-FRIT) and the backpropagation. The FRIT is a method that tunes control parameters directly by using a set of operating data. The main advantage of the FRIT is that the parameters can be determined without any mathematical model. The FRIT evaluates only a control response, whereas the E-FRIT does it as well as the difference of manipulated variable. In this paper, the E-FRIT is utilized for a learning algorithm of a neural network. Simulation examples are provided to show an effectiveness of the proposed method by simulation examples.