A strategy for diagnosing complex multiple-fault situations with a higher accuracy/cost ratio

Abstract A complex multiple-fault situation (CMF situation) occurs when several faults are present in a physical system, and affect several different entities. An example of a CMF situation is given, and four factors increasing the likelihood of these situations are identified. Their diagnosis is typically performed with a poor accuracy/cost ratio. Multiple-variable diagnosis strategy (MVDS) is a consistency-based diagnosis method that specifically addresses the CMF situations. Its originality is to avoid investigating the complete model of the system at once by dynamically updating which parts of this model should be investigated, using intermediate diagnosis results. This design provides MVDS with several advantages, which are an increase in the precision of the diagnosis and in the process’ co-operative ability, and a decrease of the computational load and of the number of measurements required. A water-distribution system is used as a case study. After its description, the way it is declared in MVDS is explained. Diagnosis processes on this system, with and without MVDS, on several fault scenarios, and using several sets of measurements, are described and compared. Conclusions are drawn about the advantages and shortfalls of using MVDS, and suggestions for further research are given.