The self- learning fuzzy proportional - integral - differential (PID) control based on neural network can bring the superiority of two control techniques

namely neural network control and fuzzy logic control

into full play. Through controller created by combination of the two said techniques

a fuzzy rule database can be established on the basis of classifying the neural networks

at the time of strengthening the function of PID control

the difficulty of establishing UK fuzzy database has also been reduced. This new type PID controller is very effective for controlling the main steam temperature with large time - delay and with non - stabilizing model in thermal power plants. Through study on system simulation

it has been shown that the self - learning PID controller based on neural network has obviously better control quality than that of the conventional PID control system

especially under conditions of operation variation

the control effectiveness being more remarkable.