Modeling and Development of Soft Sensors with Particle Filtering Approach

  • Author / Creator
  • Limitations of measurement techniques and increasingly complex chemical process render difficulties in obtaining certain critical process variables. The hardware sensor reading may have an obvious bias compared with the real value. Off-line laboratory analysis with high accuracy can only be obtained every certain period, sometimes even with time delay. Soft sensors are inferential methods that provide real-time estimation for those critical variables. This thesis deals with modeling, on-line calibration and implementation issues that are associated with soft sensor development. In chemical industries, processes are often designed to perform tasks under various operating conditions. In order to deal with modeling difficulties rendered by multiple operating conditions, the Expectation-Maximization (EM) algorithm is applied to deal with the identification problem of nonlinear parameter varying systems. The existing model is updated using the latest observation data. The particle filter based Bayesian method is proposed in this thesis to synthesize different sources of measurement information. An augmented state is constructed to deal with processes with time delay problem.

  • Subjects / Keywords
  • Graduation date
  • Type of Item
  • Degree
    Master of Science
  • DOI
  • 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.