A knowledge-based alarm processor for an energy management system

A novel station-oriented knowledge-based alarm processor prototype is presented. The knowledge-based system accesses a configuration database that contains information on both the topology of the power system and symbolic data values. Forward chaining rules are used to form hypothese when state changes occur on the power system. Backward chaining rules are used then attempt to prove each hypothesis using the current facts and the configuration database. Metalevel control rules improve system performance by selecting goals and doing context switching during the inference process. A rule-based management system is also introduced. The knowledge-based system continuously analyzes and processes SCADA (supervisory control and data acquisition) alarms in real time, presenting diagnostic messages to the power system operators without requiring any information from the operators. The prototype, written in the C programming language, resides on a stand-along computer with a data interface to the SCADA system.< >