Knowledge-based diagnostic systems for avionics

Abstract

A knowledge-based approach for diagnosis of electronic circuits in avionics systems is discussed. The knowledge-based methodology is generic and was designed for accurate and efficient identification of malfunctioning circuit components in diagnostic situations involving single-component malfunctions, multiple independent malfunctions, and multiple interacting malfunctions. The resulting methodology integrates the features of two knowledge-based techniques. Hierarchical classification (HC), a compiled knowledge approach, is used to efficiently identify single and multiple malfunctions. An approach called diagnostically focused simulation, which refines the HC results with causal reasoning, is then used to resolve whether multiple malfunctions are the result of independent component failures or whether an interaction exists that relates the malfunction behaviors. A generally-applicable shell incorporating the methodology was developed in a commercial software package and applied to the development of a diagnostic system for a black and white television circuit.<>