Implementation of Gaia methodology for multi-agent based transformer condition monitoring

Power transformer is one of the most important and expensive equipment in a power system. Building systems to monitor their real-time behaviours and diagnose their faults automatically with comprehensive knowledge-base are the key issues. This paper provides a new framework for power transformer monitoring and fault diagnosis based on a multi-agent system. The Gaia methodology is applied to clarify, simplify and standardize the design of the multi-agent system. Real-time data is sampled from a power transformer and saved into a database and it is also available to users on request. The applied rule-based reasoning for the system reduces the user´s work. Therefore, the proposed system is a generic system which can be applied to a wide range of applications, to control a particular device by applying the specific knowledge for that device added to the system. New functions can be added to the system by developing new agents.