Intelligent alarm processing and alarm elimination in a CHP installation

With combined heat and power (CHP) installations getting more and more complex, associated control systems are becoming very elaborate. For the user, this results in a higher level of difficulty for the operation of the installation. One aspect of this problem is the very high alarm rate the operator in the control room has to process. In this paper, an analysis of the alarm signals coming from a real CHP installation is made. Three different types of alarms are distinguished. For each of these types, a different approach for the possible elimination from the alarm flow is presented. The presented methods range from a simple elimination for the first type of alarms, to methods with a statistical approach for the second type. For the most complex type of alarms, a more advanced algorithm is presented: based on the formalism of fuzzy reasoning petri nets (FRPN), a tool for the diagnosis of the most likely cause of a fault is developed.