Neural network based self-tuning PID controller with Fourier transformation of temporal patterns

A neural autotuner for PID controllers that is relevant for real-time applications is presented. The neural autotuner, in an adaptive PID control scheme, processes temporal patterns (the responses of closed-loop relay feedback in the controlled system) in real time and produces updated parameters for the PID controller. Numerical experiments were carried out to test the training and generalization of the neutral autotuner with respect to different training sets, techniques of temporal signal presenting and processing and the influence of noise