control group seminar
0 views
Skip to first unread message
Arun Srikanth
Feb 12, 2010, 2:27:15 AM2/12/10
to con...@che.iitm.ac.in, controlg...@googlegroups.com, na...@iitm.ac.in, sridh...@iitm.ac.in, nkai...@iitm.ac.in, aru...@iitm.ac.in
Dear all,
Control group seminar will be held today at 4:15 pm. Venue is MSB 132. All are welcome. The talk will be given by Harshit student of Dr.Shankar narasimhan. The abstract of the talk is given below.
"Reconciliation of process data is an important pre-processing technique prior to simulation, optimization and advanced control activities. The reconciliation method requires a process model which is generally developed using first principles. For many complex processes, the development of such models is difficult and time consuming. In this work, we have proposed a method for steady state data reconciliation of nonlinear processes which does not require a functional model between variables to be specified a priori. A nonlinear model relating the variables is simultaneously developed from a data set, while reconciling the data. Kernel Principal Component Regression (KPCR) is used for identifying a nonlinear model. KPCR captures the dominant nonlinear features of the original data by transforming it to a high dimensional feature space. The extracted features are then regressed with the output variables in the original space to identify a set of regression coefficients for the model. A heat exchanger network of a crude preheat train will be used to illustrate the utility of the proposed approach."
Thanking you,
S.Arun Srikanth.
Control group seminar will be held today at 4:15 pm. Venue is MSB 132. All are welcome. The talk will be given by Harshit student of Dr.Shankar narasimhan. The abstract of the talk is given below.
"Reconciliation of process data is an important pre-processing technique prior to simulation, optimization and advanced control activities. The reconciliation method requires a process model which is generally developed using first principles. For many complex processes, the development of such models is difficult and time consuming. In this work, we have proposed a method for steady state data reconciliation of nonlinear processes which does not require a functional model between variables to be specified a priori. A nonlinear model relating the variables is simultaneously developed from a data set, while reconciling the data. Kernel Principal Component Regression (KPCR) is used for identifying a nonlinear model. KPCR captures the dominant nonlinear features of the original data by transforming it to a high dimensional feature space. The extracted features are then regressed with the output variables in the original space to identify a set of regression coefficients for the model. A heat exchanger network of a crude preheat train will be used to illustrate the utility of the proposed approach."
Thanking you,
S.Arun Srikanth.
Reply all
Reply to author
Forward
0 new messages