Building and optimizing an expert system for ATP design

A description is presented of an expert system called ADES (ATP design expert system) for the design of automatic train-protection (ATP) systems. (An ATP system is a railway signaling system consisting of a set of logic circuits that control the safe movements of trains within a railway station.) Artificial intelligence techniques proved feasible to address this particular design problem, and ADES, which uses artificial intelligence techniques, makes it possible to design good control circuits to meet operational requirements rapidly by using well-structured, explicitly represented knowledge on ATP. ADES' inference engine works as a partial evaluator for the knowledge representation language KARL which is used to write design rules. Since every circuit produced by the expert system is still expressed in KARL, the KARL interpreter has also been used as a simulator for a designed circuit. Prototypes of the inference engine and the KARL interpreter have been implemented as extended Prolog metainterpreters.<
>