Hierarchical Fault Diagnosis: Application to an Ozone Plant

A framework for diagnosis in hierarchical finite-state machines is presented and applied to an ozone plant. In this approach, the model of the system is broken into simpler substructures called D-holons. At any instant, instead of the complete system model, only the D-holons associated with the ongoing phase of operation are used for diagnosis, which reduces memory requirement. Furthermore, within the above setup, a semimodular diagnosis method is presented and used to reduce design computations. Following the proposed framework, a diagnosis system is designed for the ozone plant. It is shown that the proposed approach significantly reduces the complexity of constructing and storing the diagnosis system.