Paper
Effect of Error in Predictive Model of Model Prediction Control to Minimize Temperature Change of Plate with Varying Heat Generation
- Authors:
- Shigeki Hirasawa; Kazuya Koike; Tsuyoshi Kawanami; Katsuaki Shirai
- Abstract
- Precise temperature control to decrease temperature change of works within 0.001°C is required in semiconductor manufacturing process. We analytically and experimentally study the model predictive control to minimize temperature change at an object position in a 2-dimensional vertical plate with a varying noise-heat-generation. The noise-heat-generation is that the heating is OFF and ON every 300 s, and it causes temperature change about 3°C without control. A control-heater in the plate is controlled with the model predictive control method using the step response pattern as a dynamic predictive model. In this work, we study the effect of error in the dynamic predictive model. We found that artificial error in the dynamic predictive model causes various patterns of temperature change at the object position such as periodical error and oscillation error. Experimental result of temperature change at the object position is 0.08°C and has periodical error of 300 s period using the averaged step response pattern obtained by experiment. It is because the step response pattern obtained by experiment has random noise error. The temperature change at the object position is 0.07°C and has periodical error of 600 s period using the step response pattern obtained by the network model calculation. It is because the step response pattern obtained by the network model calculation has some modelling errors entirely.
- Keywords
- Precise Temperature Control; Heat Transfer; Process Control; Model Predictive Control; Dynamic Predictive Model
- StartPage
- 71
- EndPage
- 78
- Doi