DeepOPF-AL: Augmented Learning for Solving AC-OPF Problems with Multiple Load-Solution Mappings
This work addresses a fundamental issue in power systems optimization for grid operators, offering an incremental improvement over existing DNN methods.
The paper tackled the challenge of multiple load-solution mappings in non-convex AC-OPF problems, which can cause deep neural networks to fail; it proposed DeepOPF-AL, an augmented-learning approach that trains a DNN to learn a unique mapping from augmented inputs, achieving noticeably better optimality and similar feasibility and speedup compared to a recent DNN scheme in simulations over IEEE test cases.
The existence of multiple load-solution mappings of non-convex AC-OPF problems poses a fundamental challenge to deep neural network (DNN) schemes. As the training dataset may contain a mixture of data points corresponding to different load-solution mappings, the DNN can fail to learn a legitimate mapping and generate inferior solutions. We propose DeepOPF-AL as an augmented-learning approach to tackle this issue. The idea is to train a DNN to learn a unique mapping from an augmented input, i.e., (load, initial point), to the solution generated by an iterative OPF solver with the load and initial point as intake. We then apply the learned augmented mapping to solve AC-OPF problems much faster than conventional solvers. Simulation results over IEEE test cases show that DeepOPF-AL achieves noticeably better optimality and similar feasibility and speedup performance, as compared to a recent DNN scheme, with the same DNN size yet elevated training complexity.