A multiblock model for decentralized analysis of power system oscillations

This paper develops a novel multivariate scheme to analyze power system oscillations from real system data. The approach is based on multivariate techniques especially suited to consider the hierarchical and decentralized nature of Wide Area Monitoring Systems (WAMS), where various levels of measurement and control units are scattered along the network. It employs multiblock principal component analysis to build a statistical model based on modal signals obtained from measurements. Once the global model is constructed using decentralized local information from each part of the system, principal oscillation modes are identified along with their contribution to the different blocks.