PURPOSE:To identify a transfer function by using a neural network, to adapt PID parameters even to time deformation and a nonlinear system to which they are hardly adapted before, and to display the transfer function of a controlled system to a designer. CONSTITUTION:This method has a transfer function identification part 12 which identifies a closed loop transfer function including a PID controller 16 and a controlled system so as to determine the parameters of control elements P, I, and D automatically when a set value of a process control system including a PID control loop is varied stepwise and a PID parameter calculation processing mechanism 15 which determines the parameters of respective P, I, and D operations according to the identified closed loop transfer function. The transfer function identification part 12 constitutes the neural network which has at least one hidden layer while an input layer is time-series data of the latest (n)-sample period of process measured values and an output layer is the coefficient of the closed loop transfer function.
展开▼
