Grainger CEME

PhD student Xichen Jiang with advisor A. Domínguez-García
Rapid detection and identification of power system transmission line outages is crucial for maintaining reliability in the power grid. Although many methods exists for line outage detection, most are model based and are thus susceptible to modeling errors. We propose a measurement based statistical method to detect and identify transmission line outages in near real time. For this method, the statistical properties of electricity generation and demand are assumed to be known in addition to the power system model. To detect line outages, a statistical quickest change algorithm is applied to the voltage phase angle measurements collected from phasor measurement units (PMUs). When a line outage occurs, the statistical properties of the measured voltage phase angles change; this is detected by the algorithm. Additionally, a method to partition the power system such that the proposed algorithm for line outage detection can be applied in parallel to each area is presented. The optimal partitioning scheme is formulated as an integer program. With this algorithm, good performance can be achieved even when some of the phase angles are unobserved by the PMU. The algorithm is applied to the 3-bus system shown in Figure 33. Figure 34 shows that for an outage of line (1, 2), the stream for that tline, W(1, 2), grows more rapidly than the other streams. For a threshold of 100, W(1, 2) crosses the threshold after 17 sample points and correctly identifies the line outage. This research is funded by NSF ECCS 09 54420 CAR.

Figure 33: 3-bus system

Figure 34: Line outage detection streams