Aero-engine wavelet neural network PID control

An aero-engine proportion integration differentiation(PID) control method based on wavelet neural network on-line identification was proposed in this paper.The network structure was available with a three layers feed-forward neural network.The wavelet function was used as the activation function of the network's hidden layer.The network parameters were trained off-line.The derivative of the network's output to input was used as the engine model's derivative,and it was used to adjust the PID parameters on-line.The step response tests show that the settling time of PID control system adjusted by wavelet network is less than 2 seconds,and the steady state error is zero.The system works normally and steadily in the whole envelop.