Identification of Power Line Outages Based on PMU Measurements and Sparse Overcomplete Representation

Efficient identification of power line outages is becoming increasingly important, not only for avoiding large-scale blackouts that might be caused by rather minor power line outages, but also for many critical tasks in the smart grid, such as real-time contingency analysis, state estimation and optimal power flow control. Existing approaches to identifying power line outages are, however, either challenged by the combinatorial complexity problem, limiting these approaches to be computationally tractable for only one or two line outages, or in need of impractical assumptions such as real-time phasor angle measurements are conditionally independent and available in the whole network. Considering only a subset of the network data can be collected from phasor measurement units, and the initial number of power line outages is only a small fraction of the total number of lines, this paper reformulates the DC power flow model as a sparse overcomplete expansion and utilizes two near real-time algorithms based on greedy steps and coordinate descent iterations to identify multiple line outages at affordable complexity. Finally, the effectiveness of the proposed approach to identifying multiple line outages is validated by simulation tests on the IEEE 118 bus system.