A Systems Approach to Data Fusion for Gas Turbine Condition Health Monitoring

摘要

Current power systems diagnostic and condition monitoring systems generate information that, while accurate the majority of time, is produced without regard or access to other sources of related diagnostic information. In addition, as the control system detects faults and protects the power system when anomalous conditions arise, the information concerning that incident is not transmitted to the design personnel who can make best use of the data and subsequent assessment of the root cause(s). This leads to ambiguity in troubleshooting, requires maintenance personnel to make uninformed decisions and results in erroneous component removals and high operating costs for end users. This paper describes a systems approach to data fusion for improved gas turbine prognostics and health maintenance (PHM). The PHM system integrates sensors and model data from various diagnostic, prognostic and usage sources with information fusion algorithms to assess, to forecast, to communicate, and to protect key subsystem and component health. In addition, the communication of the key design and assessment data will allow a better understanding of the machine's behavior under known conditions and allow the designer to factor in improvements in a more rapid fashion. This will then improve the next generation of power systems and PHM techniques.