This thesis presents a novel approach for modelling complex papermaking systems via a combination of existing techniques. The project has addressed the question of modelling the process known as retention on production paper machines using a semiphysical method. This method combines both complicated physical analysis of key papermaking equipment and the use of neural networks to model the dynamic components and interactions that cannot be readily modelled via physical equations. This approach was adopted to satisfy the requirements of the sponsoring company, English China Clays International (ECCI), that the finished modelling methodology should be readily adaptable to different papermaking situations, as required. A semiphysical model was determined to meet all the criteria set by ECCI. The development of the semi physical modelling method is described through the stages of first application in a laboratory scale application (a pilot paper machine), to the final testing and validation of the technique on real data, gathered on industrial papermachines. Neural networks and their application to retention modelling are also described in this thesis, as their usage was employed in parts of the semiphysical model structure. Also, the concept of neural network error compensators are discussed. This is a novel application of a neural network to predict and then correct the modelling error of a system, thereby increasing the accuracy of the final result.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:613427 |
| Date | January 1999 |
| Creators | Rooke, Paul Edward |
| Publisher | University of Manchester |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0019 seconds