In this paper, we propose a self-tuning controller with a grey predictor for on-line process controls. The self-tuning mechanism is designed basing on a group of input-output data obtained from the process. The grey predictor in the tuning mechanism is utilized to reduce the random variation of the input-output data. The self-tuning mechanism is integrated into an internal model control, and the developed control system is applied to control the temperature distribution in a thermal barrel of plastic molding processes. From the experimental result, we conclude that the usage of the grey predictor can filter out the noise in the process and reduce the number of input-output data required in the tuning mechanism.
2001年灰色系統理論與應用學術研討會論文集=The 2001 seminar of applied grey system theory，頁C113-119