Enriching Neural Network Training Dataset to Improve Worst-Case Performance Guarantees
This work addresses the need for more reliable neural network approximations in power systems, specifically for AC-Optimal Power Flow, by reducing worst-case violations, though it is incremental as it builds on existing dataset generation methods.
The paper tackles the problem of improving worst-case performance guarantees for neural networks in power systems by proposing an algorithm that adaptively enriches the training dataset with critical datapoints, resulting in substantially reduced worst-case violations across four test systems ranging from 39 to 162 buses.
Machine learning algorithms, especially Neural Networks (NNs), are a valuable tool used to approximate non-linear relationships, like the AC-Optimal Power Flow (AC-OPF), with considerable accuracy -- and achieving a speedup of several orders of magnitude when deployed for use. Often in power systems literature, the NNs are trained with a fixed dataset generated prior to the training process. In this paper, we show that adapting the NN training dataset during training can improve the NN performance and substantially reduce its worst-case violations. This paper proposes an algorithm that identifies and enriches the training dataset with critical datapoints that reduce the worst-case violations and deliver a neural network with improved worst-case performance guarantees. We demonstrate the performance of our algorithm in four test power systems, ranging from 39-buses to 162-buses.