碩士 / 國立臺灣海洋大學 / 通訊與導航工程系 / 96 / The Global Positioning System (GPS) and inertial navigation systems (INS) have complementary operational characteristics and the synergy of both systems has been widely explored. Most of the present navigation sensor integration techniques are based on Kalman filtering estimation procedures. For obtaining optimal (minimum mean square error) estimate, the designers are required to have exact knowledge on both dynamic process and measurement noise. However, noise are unknown and varying with time, the vehicle also with difficulty guarantees continues the regular movement, therefore to establish the actual dynamic model is extremely difficult, resulting in the Kalman filtering performance degradation. The case that theoretical behavior of a filter and its actual behavior do not agree may lead to divergence problems. In system design, H∞ filter can be employed to ensure that the energy gain from the disturbances to the estimation error will not exceed a pre-specified level. The philosophy of such type of filter is designed based on the approach of linear quadratic (LQ) game theory, which is sometimes called a minimax filter due to the fact that it minimizes the worst-case performance under noise uncertainties. Results based on H∞ filtering approach may provide better performance than those based on standard Kalman filtering when the noises are non-Gaussian, or when the statistical knowledge of noise is poor. The more the uncertainty of noise knowledge is, the worse the solution of the Kalman filter becomes.
In this paper, tuning of the prescribed level of noise attenuation (r) of the H∞ filter is performed through the grey relation. The system with partial unknown structure, parameters, and characteristics is called a grey system. The grey system theory can be employed to improve the navigation accuracy performance without sufficient information or with highly nonlinear property. The grey relational analysis (GRA) uses information from the grey system to dynamically compare each factor quantitatively. This approach is based on the level of similarity and variability among all factors to establish their relation. The relational analysis suggests one approach to make prediction and decision, and generate reports that make suggestions for tuning the parameter. It also provides data to support quantification and comparison analysis. The GRA-aided H∞ filtering for GPS navigation processing is conducted and the resulting performance is discussed.
| Identifer | oai:union.ndltd.org:TW/096NTOU5300033 |
| Date | January 2008 |
| Creators | Hsin-Hsu Lin, 林信旭 |
| Contributors | Dah-Jing Jwo, 卓大靖 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 87 |
Page generated in 0.0164 seconds