ON-LINE MONITORING AND DYNAMIC FEATURE TRENDING AS A MEANS TO IMPROVE IN-SERVICE INSPECTION, MAINTENANCE, AND LONG-TERM ASSESSMENT OF SYSTEMS AND COMPONENTS

摘要

The recent publication of an international standard on vibration monitoring of reactor internal structures (IEC 61502, standard of the International Electrotech-nical Commission) concludes a long-standing development of methods for a new type of nuclear power plant instrumentation and control system. This new system differs from traditional monitoring systems, as it is aimed primarily at early failure detection. Useful information is provided to plant operators, but in particular to the inspection/maintenance personnel. Powerful modern computer technology allows―in an effective manner―correlation and spectral analysis, feature extraction and trending, alert level monitoring, and remote data/ signature transmission of dynamic process signals such as neutron noise, vibration signals, pressure noise, etc. The. signature and feature trends are stored in a central databank and form a useful reference for assessment of components and systems in case of actual incipient failure development or at the estimated end of the plant lifetime as part of the aging condition assessment measures. The standard IEC 61502 is used as an example for early failure detection and on-line condition monitoring methods based on signature analysis and feature vector monitoring in general. The reasons/background for structuring this particular standard in a mandatory and an optional part as well as the resulting consequences are explained. But emphasis is also placed on demonstration of the principles, i.e., measurement analysis and usefulness of such methods for maintenance and refurbishment strategies. The transferability of the methods is discussed when applied for other tasks, for instance, for condition/aging assessment of process instrumentation.