Knowledge representation for instructors' aids in civil flight simulators

The author investigates the suitability of fuzzy sets and rules for monitoring and assessing flight and pilot control during a simulator exercise through the intelligent interpretation of instrument readings. The emphasis is on instrument-landing-system approaches as used by civil pilots. The set and rules developed represent the knowledge required to predict critical situations, recognize the intention to correct for these, and describe the aircraft handling. A software demonstration system runs at speed comparable with real-life approaches and contains descriptions and predictions displayed linguistically. The applicability of a critiquing technique using fuzzy reasoning is investigated. This technique requires the specification of limits outside which acceptable performance is unlikely to be achieved, thus simplifying knowledge acquisition.<>