Refinement of Fuzzy Diagnosis in Decentralized Two-Level Diagnostic Structure

Abstract The novel and practicable on-line fuzzy fault isolation approach is proposed and discussed in this paper. This approach refers to proposition of diagnosing in decentralized two-level diagnostic structure. The idea of the proposed diagnostic structure is described in details as well as depicted in examples. It is particularly foreseen for applications in diagnostics of large scale industrial systems. Diagnosing in decentralized structures involves the problem of diagnosis refinement. Substantial part of the paper is devoted to clarify and solve this problem. Here, the fuzzy approach was proposed motivated by its inherent ability to handle uncertainties. Proposed approach is characterized by: better efficiency by application of parallel diagnosing, simplification of diagnostic process, providing better robustness of diagnoses compared to the centralized diagnosing. An example of industrial application of on-line decentralized diagnostics for steam-water line of power boiler was briefly discussed. Finally, the conclusions regarding main features of the decentralized two-level diagnostic structures were given.