This thesis deals with the development of Predictive Maintenance system for boiler, turbine and generator by using the Petrinet modeling method coupled with digital filtering to perform failure prediction and processing. These capabilities are very useful for health monitoring of the power plant equipments. The process consists of analysing the operating and design data of the equipment and then develop the petrinets for the fault detection system. The petrinets are then transformed into the basic building blocks. This is much simplified by use of the SIMULINK platform for the simulation of monitoring. The next step is to fix the thresholds and alarms to get appropriate maintenance signal from the system. The major requirement is thorough study of the operating plant systems. It entails the in depth knowledge of the plant data as the information regarding the excursions of the concerned parameters during the normal operation and also the changes during occurrence of the problem would be essential for setting of proper thresholds.
With the availability of cheap and high speed computing power in the modern microcomputers the implementation of the system would be cost effective. The system developed could be effectively used for existing power plants with retrofitted DAS/DCS.

Keywords: Predictive Maintenance, Petrinets, DAS, BFP, CEP,BTG.