Anticipating contingengies in power grids using fast neural net screening

We address the problem of maintaining high voltage power transmission in security at all time. This requires that power flowing through all remain below a certain nominal thermal limit above which lines might, break or cause other damages. Current practices include enforcing the "N-1" reliability criterion, namely anticipating exceeding of limit for any eventual single line disconnection (whatever its cause be) by running a slow, but accurate, physical grid simulator. New frameworks are calling for a probabilistic risk based security and are in need of new methods to assess the risk. To tackle this assessment, we address in this paper the problem of rapidly ranking order contingencies including all pairs of line disconnections, to prioritize simulations. We present a novel method based on neural, which ranks "N-1" and "N-2" contingencies in decreasing order of severity. We demonstrate on a classical benchmark problem that the risk of contingencies decreases dramatically compared to considering all "N-1" cases, at no additional computational cost. We evaluate that method scales up to power grids of the size of the French high voltage grid (over 1000 power lines).