DISTRIBUTION-FREE POSSIBILITY MODELLING OF POOR SENSOR INFORMATION |
|---|
| G. Mauris, V. Lasserre, L. Foulloy |
- Abstract:
- At the application level, it is important to be able to define around the measurement result an interval which will contain an important part of the distribution of the measured values, that is, a confidence interval. When the sensor uncertainty is represented by a probability distribution, the confidence intervals can be easily deduced from it. But when the probability distribution cannot be identified due to poor sensor information, a more generalised representation must be used. To obtain confidence intervals in such a situation, available probabilistic methods are essentially the Bienayme-Chebychev and the Camp-Meidel inequalities. In this paper, after having recalled these methods, alternative approaches based on the possibility theory are considered. Distribution-free possibility distribution building based on the sets of all confidence intervals is proposed. According to the knowledge of uncertainty that is available, i.e. the range or the standard deviation of the measures, triangular and truncated triangular possibility distributions are respectively considered. These different possibility distributions which are fuzzy sets with uncertainty semantics are then compared in terms of the information provided.
- Keywords:
- measurement uncertainty, possibility theory, probability theory
- Download:
- IMEKO-WC-2000-TC7-P194.pdf
- DOI:
- -
- Event details
- Event name:
- XVI IMEKO World Congress
- Title:
Measurement - Supports Science - Improves Technology - Protects Environment ... and Provides Employment - Now and in the Future
- Place:
- Vienna, AUSTRIA
- Time:
- 25 September 2000 - 28 September 2000