New Algorithm Remotely Locates Faults In The Power Grid

The power grid is aging and, according to many, is experiencing more faults than ever — fault that could lead to extended power outages and equipment damage.

To help with this potential damage, a team of researchers from Binghamton University have proved that the Singular Spectrum Analysis (SSA) algorithm may be the best tool to help remotely detect and locate power grid faults. The SSA algorithm is a nonparametric spectral estimation method that combines elements of classical time series analysis,multivariate statistics, multivariate geometry, dynamical systems and signal processing. The algorithm, with roots back into the 18th century, can assist in the decomposition of time series into a sum of components that each having a meaningful interpretation.

“Theoretically, the SSA algorithm is an optimal approach for accurate and quick detection. However, it has not been adopted in real-world engineering applications. We adapted and improved the algorithm for the new application in power grid areas,” said Yu Chen, associate professor of electrical and computer Engineering at Binghamton University, and co-author of the paper with Zekun Yang, Ning Zhou and Aleksey Polunchenko.

Typically, multiple complete circuits (a grid) keeps electricity flowing even when one wire goes down.

This provides stability but is complex and susceptible to vulnerabilities — tree limbs can take out wires in a windstorm or even hackers can break in and change the flow of electricity subtly, resulting in major negative effects.

Currently, there are formulas in existence, such as the Event Start Time (EST) algorithm, which calculates differing arrival times of power changes in different geographic locations, to determine where faults are in the grid.

However, the Binghamton team used simulation data generated by the Power System Tool box to prove that the SSA algorithm is faster and more reliable when it comes to finding changes in the power grid from generator or transmission line problems. SSA could even be used to predict future problems.

“At the current stage, the algorithm can only detect and locate problems, and it cannot predict future problems,” said Zhou. “It laid a solid foundation for the next step: prediction. Being able to detect subtle changes in the power grid promptly, our approach has the potential to predict future problems by including a power system model.”

Although the team has confirmed that the SSA is quite effective, it will still need to fine-tune it in order to include more ways of gathering accurate geolocations of problems, more simulation testing and real-world data collection to “validate the algorithm and polish it to cope with more realistic scenarios.”