Use this URL to cite or link to this record in EThOS:
Title: Artificial neural network-based control for process tomography applications
Author: Benchebra, Dalil
ISNI:       0000 0004 2671 7720
Awarding Body: The Manchester Metropolitan University
Current Institution: Manchester Metropolitan University
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Electronic Capacitance Tomography systems have proved to be extremely useful in non-invasive measurement for industrial applications. While considerable research effort has been focused on the development and refinement of measurement techniques based on ECT, use of ECT for realtime control has not attracted the same extent of research effort. This work demonstrated that a novel combination of ECT systems and Artificial Neural Networks (ANNs) can be used to control of highly nonlinear industrial systems continuously as demonstrated by the implementation of a neural network-based inverse controller for the MMU laboratory flow. rig conveying the polypropylene pellets. The nature of the flow ofpneumatically conveyed pellets is highly nonlinear which tends to lead to the formation of dunes, necessitating an increase of air velocity to the maximum to clear the dunes, and hence requiring large control energy. If the air velocity can be controlled such that the pellet flow is maintained at a constant rate without the build up of dunes, energy usage associated with such processes can be considerably reduced. One of the main problems in the control of the pneumatic pellet flow system is the difficulty in building good models of the nonlinear dynamics of the system. In this work, ANNs are used to initially build and validate a model of the forward dynamics of the system and then to develop an inverse model of the plant. This inverse model is implemented as a Controller to maintain constant pellet flow and to clear dunes as quickly as possible. Results are obtained from a laboratory flow rig interfaced to a Virtual Instrument Tomographic Measurement System and controlled using dedicated hardware with software implemented in LabView and Matlab. The NN-based controller was highly effective in maintaining a steady pellet flow over long durations of time even in the presence of mass flow disturbances. The results presented in this work showed that a NN-based controller can eliminate energy wastage by automatically clearing dunes as and when they form while maintaining the air velocity at a minimum value necessary to keep the pellet flow homogeneous. Serial and Parallel ECT systems were used in the imaging and control experiments. The additional information obtained by the high imaging rates of the parallel ECT system was used to improve the performance of the controller for long term operations. Hierarchical Self-Organising Maps were also shown to be highly effective in improving the accuracy of the images obtained using the standard Linear Back Projection algorithm for application in ECT-based systems.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available