- 415 views
- 957 downloads
Data-driven methods for near infrared spectroscopy modeling
-
- Author / Creator
- Chen, Mulang
-
Time consuming offline laboratory analysis and high cost hardware measurement techniques render difficulties in obtaining the important quality variables in real time application. Near-infrared (NIR) spectroscopy is widely used as a process analytical tool (PAT) in chemical processes, providing online estimation of the target properties which are often obtained by lab analysis. This thesis focuses on the model building, model structure (wavelength) selection and online model update for NIR applications.
Time varying issue is solved by applying recursive adaptation methods and a novel recursive wavelength selection algorithm is proposed to adapt the model structure during online phase. The Just-in-time (JIT) modeling approach is adopted to model the nonlinear relationships between spectra and properties. A similarity criterion that utilizes input-output information is developed to search for most relevant samples from the database. Finally, the recursive algorithm and locally weighted algorithm are synthesized into the JIT framework in order to deal with both time varying and non-linearity issues of the process.
-
- Subjects / Keywords
-
- Graduation date
- Fall 2013
-
- Type of Item
- Thesis
-
- Degree
- Master of Science
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.