Use of data mining techniques in the performance monitoring and optimisation of a thermal power plant

The article describes research currently being carried out by the Control of Power Systems Group at the Queen's University of Belfast into the application of data mining to the performance monitoring and optimisation of the steam generation systems in thermal power plants. This work is being carried out in conjunction with Premier Power plc which owns and operates Ballylumford Power Station near Larne in Northern Ireland; this station consists of 3/spl times/120 MW and 3/spl times/200 MW gas/oil fired generating units, plus 2/spl times/60 MW gas/oil turbines. The main components of a steam generation system consist of an oil/gas fired boiler, a turbine and a condenser. Although the operation of these is conceptually simple, the components are extremely complicated and due to the nature of the processes involved in steam generation, they are prone to degradation and failure. This can lead to a reduction in the thermal efficiency of the plant, increases in plant emissions and the possibility of unscheduled power outages. The aim of the research is twofold: to develop models of the plant over the full range of operating conditions; and to develop and implement a system which will use the models to determine the condition of the plant accurately, and which will be able to make operational suggestions to engineers/operators to rectify any deviations detected. The models are to be created by data mining on the large database of archived plant data which Premier Power has made available to Queen's University.