Advanced Alarm-Management Post-Processor Prototype for a Fossil Power Station

Abstract

During a power station disturbance, the data processing and monitoring system (DPMS) may present an overwhelming number of alarms (200-400) to the power station operators within a short period of time. This may lead to operator confusion and makes problem analysis and decisions difficult. An advanced post-processor Connected to the existing DPMS is being developed at IEC. In the present stage of this project our aim is to develop a model for dynamic alarm management and adaptive operator guidance that will permit the verification of the implementation methodology. The hierarchical structure of the power station is mapping the different equipment features and technological processes, to the structure used by our alarm-management system. On the basis of this double hierarchy, all the DPMS points were grouped and each DPMS point labeled with its alarm priority level (i.e., urgent alarm, alarm, warning, not an alarm). Cause-consequence graphs were constructed. The nodes of the graphs were the selected points of the DPMS. According to our methodological proposal the cause-consequence connections between the selected points of the DPMS may be constructed on the basis of the ready made ’flow diagram - process and instrument diagrams’, ’control logic diagrams’ and ’control and instrument diagrams’ of the power station documentation. Our prototype includes some limited parts of the power station such as deareator level control, one of the turbine driven boiler feed-water pumps and cooling system of the unit main transformer. About 100 rules were generated from the cause-consequence graphs. Some 200 inputs were simulated off-line. The user interface and the inference machine were built by using the KAPPA PC (IntelliCorp, Inc.) expert system shell. This prototype is being tested now in the Training Center of the Operation Division.