Inertia-Constrained Generation Scheduling: Sample Selection, Learning-Embedded Optimization Modeling, and Computational Enhancement

Mingjian Tuo, Fan Jiang, Xingpeng Li, Pascal Van Hentenryck. IEEE Transactions on Power Systems, 2025.
Data
FCUC-24bus-Sys_PSSE-TimeDmn-Data
Authors in the RPG Lab
(might be unordered here; check the citation below)
Fan Jiang
Mingjian Tuo
Xingpeng Li

Abstract

Day-ahead generation scheduling is conducted by solving securi-ty-constrained unit commitment (SCUC) problem. Fast-growing inverter-based resources dramatically reduce grid inertia, com-promising system dynamic stability. Traditional SCUC (T-SCUC), without any inertia requirements, may no longer be effective for renewables-dominated grids. To address this, we propose the active linearized sparse neural network-embedded SCUC (ALSNN-SCUC) model, utilizing machine learning (ML) to incorporate system dynamic performance. A multi-output deep neural network (DNN) model is trained offline on strategi-cally-selected data samples to accurately predict frequency sta-bility metrics: locational RoCoF and frequency nadir. Struc-tured sparsity and active ReLU linearization are implemented to prune redundant DNN neurons, significantly reducing its size while ensuring prediction accuracy even at high sparsity levels. By embedding this ML-based frequency stability predictor into SCUC as constraints, the proposed ALSNN-SCUC model mini-mizes its computational complexity while ensuring frequency stability following G-1 contingency. Case studies show the pro-posed ALSNN-SCUC can enforce pre-specified frequency re-quirements without being overly conservative, outperforming five benchmark models including T-SCUC, two physics-based SCUC, and two ML-based SCUC. The proposed sparsification and active linearization strategies can reduce the DNN-SCUC computing time by over 95% for both IEEE 24-bus and 118-bus systems, demonstrating the effectiveness and scalability of the proposed ALSNN-SCUC model.

Index Terms

Deep learning, Frequency deviation, Frequency stability, Line-arization, Low-inertia power systems, Sparse neural network, Rate of change of frequency, Unit commitment

Cite this paper:

Mingjian Tuo, Fan Jiang, Xingpeng Li, and Pascal Van Hentenryck, “Inertia-Constrained Generation Scheduling: Sample Selection, Learning-Embedded Optimization Modeling, and Computational Enhancement”, IEEE Transactions on Power Systems, Dec. 2025.