Study of Neural Network PID Control in Variable-frequency Air-conditioning System

This paper aims at the control of variable-frequency air-conditioning system that has characteristic as large inertia and pure lag. The neural network PID control in the variable-frequency air-conditioning system is introduced and simulated. In the learning algorithm of neural network PID controller, the output of system is needed to tune the weights of neural network while it is difficult to obtain. So the output of system is predicted through the algorithm of nonlinear that adopts the neural network configure. Through simulation and optimization, it is found that the neural network PID control has the capability of self-study and self-adaptation. However, the neural network PID control system sometimes has the static error. To eliminate the static error, the hybrid control of neural network PID and conventional PID is applied to the variable-frequency air-conditioning system. The hybrid control is simulated to compare the performance of changed parameters of model. The simulation finds that the hybrid control of neural network PID and PID has both the advantages of neural network and PID, such as self-studying and self-adapting and obtain faster response and better performance.