An auto-tuning Grey-Neuro-PID controller
Shuen-Jeng Lin, Chia-Chang Tong, Neng-Kai Yang · 2007
In this paper, we propose to add Grey prediction model GM(1,2) into the self-tuning Neuro-PID controller based on radial basis function (RBF) algorithm to improve the performance of the controller. Initially, the prediction of system output by the simple GM(1,2) model is added to the RBF algorithm as one of the inputs to enhance the performance of RBF neural network system identifier. The output of this GM(1,2)-RBF on-line learning system model is subsequently used to establish a set of updating algorithms for the gains of self-tuning PID controller. The detailed description of the proposed system structure and the design algorithm is given in this paper. The proposed auto-tuning PID controller via GM(1,2)-RBF algorithm is put into tests by Matlab simulations and motor speed control experiments by using LabVIEW. The system responses of self-tuning PID controller based on GM(1,2)-RBF and RBF are compared. Both simulations and motor test results confirm that the proposed self-tuning PID controller based on GM(1,2)-RBF performs better than the one based on RBF.