SENSOR FAULT DETECTION IN THERMAL POWER PLANTS BASED ON ASSOCIATION RULE

A new fault detection approach for sensor in thermal power plants is proposed, which is based on the association rule algorithm of the data mining in order to overcome the bad effect of the diagnosis result as modeling error and make full use of the running data. The improved association rule algorithm of data mining takes full advantage of large redundant information in the process of thermal power plants according to the multi-sensor character and the different running phases. The result demonstrates that the developed algorithm can be used to fault detection of different kinds of sensors and improves the calculation speed with association rules' evaluations and simulation.