Asymptotically stable neural networks for identification of nonlinear dynamic systems
| dc.contributor.author | Jubien, Chris Michael | en_US |
| dc.date.accessioned | 2024-08-14T17:55:55Z | |
| dc.date.available | 2024-08-14T17:55:55Z | |
| dc.date.copyright | 1993 | en_US |
| dc.date.issued | 1993 | |
| dc.degree.department | Department of Electrical and Computer Engineering | |
| dc.degree.level | Master of Applied Science M.A.Sc. | en |
| dc.description.abstract | This thesis deals with the use of stable neural networks for identification of nonlinear dynamic systems. Neural Networks have been in increasing use. Lately, they have been applied to identification of nonlinear systems. Neural networks have the advantage that they are extremely flexible and, once trained, provide accurate and fast modeling of complex systems. Proofs are available which show that neural networks can implement systems of arbitrary complexity. Here, a novel approach to system identification is used. A class of recurrent neural networks which have been proven to be asymptotically stable are the basis for identification. Equations for training these neural networks to model the complex behaviour that nonlinear systems exhibit are developed. Computer simulations are used to test the theory. In a simulated system, the neural network is found to provide good modeling. The equations are also applied to train a neural network to model the dynamic behaviour of a boat and a PUMA-560 two-link robot. | en |
| dc.format.extent | 128 pages | |
| dc.identifier.uri | https://hdl.handle.net/1828/18347 | |
| dc.rights | Available to the World Wide Web | en_US |
| dc.subject | UN SDG 17: Partnerships | en |
| dc.title | Asymptotically stable neural networks for identification of nonlinear dynamic systems | en_US |
| dc.type | Thesis | en_US |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- JUBIEN_Chris_Michael_MSC_1993_557700.pdf
- Size:
- 26.71 MB
- Format:
- Adobe Portable Document Format