Composite single neural PID controller based on fuzzy self-tuning gain and RBF network identification

Nonlinear, time-varying and uncertainty matter is an important issue in order to improve conventional PID control technology due to the rapid development of control technologies for a wide range of industries. Artificial neural PID controller such as the single neural PID controlling is a current interesting research topic because of its potential ability of auto-tuning by simulating the human brain functions. In the conventional single neural network control method, it is difficult to realize a high learning speed and online tuning PID gains. A new approach to improve the single neuron PID control technology is to combine fuzzy controlling with RBF neural network. Such a composite neuron adaptive PID controller with fuzzy self-tuning gain takes advantages of online identification and self-tuning gain thus to speed up the learning process and rationally adjust the gain coefficient, providing a much improved solution to reduce overshoot and to speed up self-learning. In this article, we show our two simulations including a second-order motion control and time-delay model to verify the effectiveness of the composite neuron adaptive PID controller method. We conclude that the adaptive fuzzy gain compound neural network PID control is superior in controlling nonlinear and time-delay objects over the conventional method. The improved control algorithm has excellent tracking performance and accuracy to meet the requirements for the next generation of intelligent controllers.