Performance characteristics of an improved single neuron PID controller using additional error of an inversed control signal

Abstract

A novel single neuron adaptive PID controller system is proposed by utilizing a new quadratic error performance function. In this new quadratic of error performance function, an additional error of thereference control signal between the input of the plant and the inversed of the plant output is added to the conventional input error of the single neuron controller to update the neuron weights in its learning mechanism. To accomplish the new quadratic error function, an inverse neural network is necessary put into the structure of the control system, for providing the inversed plant output as the feedback of the control signal. By using this new quadratic error performance function, the single neuron PID controller system is evaluated and compare with that of the conventional single neuron PID system, in terms of the rise time, the settling time and the overshoot of the system. The simulation results show that the convergence speed of the new adaptive single neuron PID system is increased and the control performance is greatly improve, outperformed the classical single neuron adaptive PID controller, especially in its strong robustness and better self-adaptation.