Currently process industry faces a paradoxical situation. On the one hand there is the urgent need to optimise the performance of processes by increasing throughput, decreasing operating costs while increasing the product quality. On the other hand there are only few specialists in industry who are able to develop and apply appropriate control strategies for the increasingly complex processes in the process industry. Generally, these specialists work in research and development departments necessitating a considerable amount of time to develop sophisticated solutions for specific processes. However, in the process industry control design and fine-tuning are mostly done by practitioners more than by specialists, directly at the process and in a minimum of time. Within this commissioning phase the process is assembled and set into operation, often with suboptimally turned controllers. Efforts have been undertaken to support these commissioners doing their tasks, and for single variable processes practically applicable methods have been developed. Nevertheless, for more complex processes the generation of mathematical process models as an appropriate base for control system design still is a major problem in practice. The subject of this work is the development of a structured approach to identification techniques for the analysis of industrial processes that enables industrial users with limited control engineering knowledge to design process models suitable for the design of industrial controllers. This latter aspect has been addressed within the collaborative research project between the University of Glamorgan and the Fachhochschule Hannover, of which the work presented in this thesis is a substantial part. Therefore, an industrially suitable scheme for computer aided control system design (CACSD) has firstly been developed in agreement with industrial users in order to set the frame for the research project. This scheme has been based on simple block-oriented model structures composed from nonlinear static and linear dynamic characteristics. The scheme is simple in use and intuitive to understand and follow. Therefore, it can be directly applied also by inexperienced engineers, who look for quick and efficient solutions as a basis even for nonlinear controller design. Beyond this a standardised identification procedure for nonlinear processes has been elaborated in order to provide process models fitting to the CACSD scheme. This standardised identification procedure has been equipped with two improved algorithms. For the approximation of even multidimensional static characteristics a capable method has been developed necessitating neither apriori information nor user interaction. For the identification of discrete-time linear dynamic models a two-step identification method has been improved by a numerically efficient least squares estimator that allows the parallel estimation of a set of model structures, which is evaluated automatically. For the validation of the proposed approach and the developed methods a prototype identification tool has been programmed, which also lays the ground for the integration of the whole CACSD scheme into a block-oriented simulation environment.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:311908 |
| Date | January 1999 |
| Creators | KoĢrner, Steffen |
| Publisher | University of South Wales |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://pure.southwales.ac.uk/en/studentthesis/a-structured-approach-to-identification-techniques-for-the-analysis-of-industrial-processes(5a5a8c62-2222-4476-86cd-7811effb3d70).html |
Page generated in 0.0021 seconds