A multi-level diagnosis methodology for complex systems

Abstract

A multi-level approach is discussed for robust diagnosis of complex continuous-valued engineering systems. The focus of the research was on component-oriented diagnosis. The diagnosis system consists of two modules: an associational module, based on heuristic knowledge gained from manuals and expert mechanics, which generates partial decision trees for mechanics to study and conduct further tests, and a model-based diagnosis module that incorporates schematic, functional, and behavioral knowledge for fault isolation and measurement selection. The architecture of the diagnosis system is presented. A key component of this system is the global controller that coordinates the activities between the two diagnostic modules. A successful prototype for diagnosing problems in the pneumatic system of a DC-10 aircraft has been implemented.<>