Fault Management as a Controller

Abstract

Fault Management (FM) architectures use control theory terms such as sense, act, state and feedback loop to describe their FM functions. This is often the extent of the comparison to closed-loop feedback control theory, as FM architectures do not leverage the linear algebra equations of modern control theory. It is natural to understand why this occurs. Control theory addresses numeric control of continuous setpoints while FM control addresses symbolic control of discrete setpoints. A closer examination of the mathematical equations of modern control theory shows it to have real benefits for FM control. It provides a canonical set of control law equations that allow alternative FM implementations to be easily compared. It provides deterministic knowledge acquisition templates for the classes of knowledge needed by FM control applications. It can scale to extremely large systems. This paper introduces the canonical equations of modern control theory, illustrates their applicability using examples from both nominal and FM control and provides a procedure for users to extract the required knowledge to support their own FM solutions.