Search

A method for auto-tuning of PID temperature control parameters (especially, of P parameter) has been investigated by utilizing microcomputer PC9801F2 as a digital controller. The presented tuning procedure, the basic concept of which lies in introducing squared control-areas of the small amplitude step response as a performance index for process parameter optimization and minimizing the characteristic areas by the gradient descent method proportional to thier relative differences, can be said to be a brief and convenient technique since there is no need for complex identification system for the estimation of a process. The optimization of the PID parameters, which is based on the optimal regulator design method, is also found to be effective for process control, though the problem of the saturation of manipulated variables must be taken into consideration. The MRAS augmented by the integral action can be expected as a practical tuning technique in the very point that the present method requires no explicit knowledge about the process transfer function.

A method for auto-tuning of PID temperature control parameters (especially, of P parameter) has been investigated by utilizing microcomputer PC9801F2 as a digital controller. The presented tuning procedure, the basic concept of which lies in introducing squared control-areas of the small amplitude step response as a performance index for process parameter optimization and minimizing the characteristic areas by the gradient descent method proportional to thier relative differences, can be said to be a brief and convenient technique since there is no need for complex identification system for the estimation of a process. The optimization of the PID parameters, which is based on the optimal regulator design method, is also found to be effective for process control, though the problem of the saturation of manipulated variables must be taken into consideration. The MRAS augmented by the integral action can be expected as a practical tuning technique in the very point that the present method requires no explicit knowledge about the process transfer function.