Power Grid Behavioral Patterns and Risks of Generalization in Applied Machine Learning
This addresses the problem of impractical ML applications in power grids for researchers and practitioners, by emphasizing domain-specific risks, though it is incremental in focusing on evaluation rather than new solutions.
The paper investigates real-world power grid data to identify spatiotemporal patterns in load, generation, and topology, and evaluates how ignoring these patterns in machine learning models leads to infeasible or meaningless predictions, highlighting risks in generalization.
Recent years have seen a rich literature of data-driven approaches designed for power grid applications. However, insufficient consideration of domain knowledge can impose a high risk to the practicality of the methods. Specifically, ignoring the grid-specific spatiotemporal patterns (in load, generation, and topology, etc.) can lead to outputting infeasible, unrealizable, or completely meaningless predictions on new inputs. To address this concern, this paper investigates real-world operational data to provide insights into power grid behavioral patterns, including the time-varying topology, load, and generation, as well as the spatial differences (in peak hours, diverse styles) between individual loads and generations. Then based on these observations, we evaluate the generalization risks in some existing ML works causedby ignoring these grid-specific patterns in model design and training.