SIMPREBAL: An Expert System for Real-Time Fault Diagnosis of Hydrogenerators Machinery

This paper proposes an expert system to aid plant maintainers and operators personnel for solving hydroelectric equipments troubleshootings. The expert system was implemented into intelligent maintenance system called SIMPREBAL (Predictive Maintenance System of Balbina). The SIMPREBAL knowledge base, the architecture and the inference machine are presented in detail. The knowledge base is based on experts empirical knowledge, work orders, manuals, technical documents and operation procedures. The predictive maintenance system architecture is based on the OSA-CBM framework that has seven layers. The software application has been successfully implemented in client-server computational framework. The data acquisition and intelligent processing tasks were develop in the server side and the user interface in the client side. The intelligent processing task is an expert system that use JESS inference machine. During two years, the SIMPREBAL has been used for monitoring and diagnosing hydro generators machinery malfunctions. The industrial application of the SIMPREBAL proved its high reliability and accuracy. Finally, satisfactory fault diagnostics have been verified using maintenance indicators before and after the SIMPREBAL installation in the hydroelectric power plant. These valuable results are been used in the decision support layer to pre-schedule maintenance work, reduce inventory costs for spare parts and minimize the risk of catastrophic failure.