PLS Model Based Control for the Tennessee Eastman Process

• Main Authors: YANG, HSI-HAUI, 楊璽懷
• Other Authors: WANG, GUO-BIN
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/39176953630994247422

碩士 === 長庚大學 === 化工與材料工程研究所 === 93 === Most of the chemical processes belong to multivariate systems. In practical situation they will face with the interactive problems. Although using PID with RGA pairing and the decoupler can handle these problems, the control results are not good enough, and the tuning algorithms often depend on operators’ experience. In the real process MPC has already successfully used in many MIMO systems. It is known that MPC methods can overcome the interactive nature and especially handle the system constrains. But so far the MPC is not popular in manufacturing plant. This work uses PLS model based PID control to deal with the multivariable systems. This PLS method is suited in the systems which have strong interactive nature. It can decompose the interactive nature and also automatically select the control pairs in the latent space without using RGA. The main purpose of this study is to apply PIDPLS algorithm and DMC approach to treating benchmark Tennessee Eastman Process. The simulation results show that PIDPLS algorithm can easily handle the systems with strong interaction and DMC approach surely provides good control performance.