Optimizing Power Grid Network Robustness

Inspiration

There has been increased funding and attention towards securing the power grid and its many inefficiencies U.S. Presidential Budget Plan Substations targeted for extremist attacks Increasing power demands and shortages Power grid failures have increased primarily because of weather related events

What it does

The minimum amount of data that a nation/region can provide for power grid metrics is purely connectivity of certain nodes with negligible distances estimations. First we need to import the data and record multiple rough centrality metrics: EigenVector Centrality, Katz Centrality, and Degree Centrality. Eigenvector Centrality can be mapped to a QUBO problem. Check the intersections between each of the largest 100 and choose the most corroborated.
Then we need to map it to a maximum clique cover problem between the set of all designated power generators. It is formulated than that all nodes excluded from the maximum clique are high priority nodes to add links to. We then computed the average degree of power generators to other power generators Compute the smallest distances between the excluded nodes and connected nodes and return the edges (or transmission lines between generators nodes that are to be placed). Those are where new transmission lines must be made

How we built it

It was entirely a programming project, written in Python (Google Colab environment) with NetworkX and DWave Libraries as the main tools.

Challenges we ran into

The runtime of the tests. They were extremely slow to run since this is a very computationally heavy project, so adjusting constraints in the time allocation given was a definite challenge.