Decentralized PMU Placements in a Dynamic Programming Approach

In order to fit in latency requirements for wide-area monitoring, protect and control (WAMPAC) applications, a decentralized architecture collaborated with measurement systems and cyber-physical systems is considered. To this end, a tree decomposition (TD) technique is utilized to decompose power grids. Many difficult problems can thus be solved by knowing local information only and/or exchanged knowledge with external regions. System-wide observability are geometrically analyzed by studying the relative power dominating set (PDS) problem in graph theory under the decentralized architecture. The hardness of subproblems determined by the maximal tree node size of a TD, denoted as parameter $k$, is also discussed. Extensive simulations are conducted on several IEEE testbed systems to demonstrate the effectiveness of the proposed formulation.