The characteristic of a fuzzy controller is to use a knowledge expressed in natural language in the forms of expert rules to calculate the value of the control starting from the numerical information from the sensors. The implementation of such a pilot requires the interpretation of the numerical field symbolically and reciprocally. This is carried out by the use of fuzzy assemblies. The associated theory allows starting from consideration of precise measurement, the whole of the stated expert rules, and choice of suitable operators to deduce a value from control. It results in a particular structure of the fuzzy controllers, whose main characteristics are the subject of the first part of the presentation. The implementation on the example of the longitudinal pilot control of an aircraft is then presented. Two approaches were adopted: the first uses the rules of manual piloting formulated in natural language by the 'experts' who are the pilots. The second uses the laws resulting from a conventional autopilot to build rules of piloting. The results are presented in the perspective from a comparison between the two types of controllers, conventional or fuzzy.
Evaluation des Techniques de Controle Flou pour le Pilotage d'Avion (Evaluation of the Techniques of Fuzzy Control for the Piloting an Aircraft)
1995
10 pages
Report
Keine Angabe
Englisch
Evaluation des techniques de controle flou pour le pilotage d'avion
British Library Conference Proceedings | 1995
|