Application of intelligent agents in power industry: Promises and complex issues

Abstract

Agent-based systems have emerged in engineering applications dealing with increased uncertainty and dynamic complex environments. This paper describes how goal oriented software components have been used for data classification of partial discharge (PD) signals produced from a Gas Insulated Substation (GIS) chamber, for PD defects' diagnosis, which required data collection, data processing, classification and finally information extraction. The intelligent decision support required knowing how information from the data set is produced, measured and interpreted, in order meaningful conclusions to be produced from the agents. The knowledge repositories given contained the acquired data, while agents could be used for data analysis, integration and understanding of knowledge. Different issues will be discussed that appeared during the design and development of the multi-agent system, aiming to be usable by other similar agent-based applications and promote the networks of knowledge, with agents producing information for the decision making of diagnosis tasks in engineering applications.