A Belief–Desire–Intention Multi-agent Architecture for Efficient Power Plant Disturbance Analysis

The amount of data produced in power system monitoring networks continues to grow indistinctly, specially in the context of smart grids, demanding new frameworks to extract information of the databases. Given the increasing number of propositions of multi-agent systems (MAS) for smart grids control, operation and restoration, these frameworks must be compatible with agent concepts. Although multi-agent systems for processing oscillographic data have been proposed, no proper method was developed to demonstrate computational benefits in the case of high volume of data. In order to address this matter, this paper proposes a set of guidelines for the development of MAS. Based on the proposed architecture, a prototype was developed, and a set of tests were performed on the oscillographic database of the Itaipu power plant, which includes the evaluation of oscillography classification accuracy, the analysis of parallel processing and the automatic analysis of a set of real disturbances. The results showed that the architecture is suitable for disturbance analysis automation, fulfilling the functional requirements and reducing processing time when compared with single thread computation.