Model-Based Fault Diagnosis in Information Poor Processes

A theory of model-based fault diagnosis is proposed which is suitable for non-linear plants that are information poor. That is, there are a bare minimum of sensors available to operate the process without recourse to analytical redundancy, the sensors output at frequencies which are likely to be low, relative to the dynamics of the plant, and there is considerable uncertainty surrounding any mathematical models that are available. Other approaches are likely to be more suitable for information rich plants. However, it should be of, at least, philosophical interest to the diagnostician who assumes that he is dealing with such a plant, if only because it should lead him to question whether his plant actually satisfies criteria necessary to support this assumption