Online Model-based Diagnosis for Multiple, Intermittent and Interaction Faults

This paper extends model-based diagnosis (MBD) (Reiter, 1987; de Kleer and Williams, 1987) to systems which convert, move and process materials or objects. Examples of such systems are printers, refineries, manufacturing lines and food processing plants. Such plants present two challenges to model-based diagnosis: (1) the plant may process with very high speed while handling multiple objects in parallel such that retaining full details of behavior of all past objects is impractical, and (2) complex multi-way interactions can occur among components operating on the same object. We address the first challenge by synopsizing past behavior and the current knowledge in a data structure of linear size in the number of components in the system. The second challenge is addressed by introducing the notion of interaction fault. An interaction fault is present if a set of components operating on the same object, damage the object even though each component alone produces non noticeable damage. Introducing interaction faults is much simpler than introducing fine-grained models of component-object interactions. We demonstrate the approach on a highly redundant printer.