Structure-Aware Commitment Reduction for Network-Constrained Unit Commitment with Solver-Preserving Guarantees

Guangwen Wang; Jiaqi Wu; Yang Weng; Baosen Zhang

cs.LGApr 3, 2026

Structure-Aware Commitment Reduction for Network-Constrained Unit Commitment with Solver-Preserving Guarantees

Authors: Guangwen Wang, Jiaqi Wu, Yang Weng, Baosen Zhang

Organizations: School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ · Department of Electrical and Computer Engineering, University of Washington, Seattle, WA

Abstract

The growing number of individual generating units, hybrid resources, and security constraints has significantly increased the computational burden of network-constrained unit commitment (UC), where most solution time is spent exploring branch-and-bound trees over unit-hour binary variables. To reduce this combinatorial burden, recent approaches have explored learning-based guidance to assist commitment decisions. However, directly using tools such as large language models (LLMs) to predict full commitment schedules is unreliable, as infeasible or inconsistent binary decisions can violate inter-temporal constraints and degrade economic optimality. This paper proposes a solver-compatible dimensionality reduction framework for UC that exploits structural regularities in commitment decisions. Instead of generating complete schedules, the framework identifies a sparse subset of structurally stable commitment binaries to fix prior to optimization. One implementation uses an LLM to select these variables. The LLM does not replace the optimization process but provides partial variable restriction, while all constraints and remaining decisions are handled by the original MILP solver, which continues to enforce network, ramping, reserve, and security constraints. We formally show that the masked problem defines a reduced feasible region of the original UC model, thereby preserving feasibility and enabling solver-certified optimality within the restricted space. Experiments on IEEE 57-bus, RTS 73-bus, IEEE 118-bus, and augmented large-scale cases, including security-constrained variants, demonstrate consistent reductions in branch-and-bound nodes and solution time, achieving order-of-magnitude speedups on high-complexity instances while maintaining near-optimal objective values.

Figures & tables

Fig. 1 : Workflow of the proposed solver-preserving commitment variable reduction method. An LLM generates a sparse commitment restriction set, a feasibility screening step removes inconsistent selections, and the resulting reduced UC problem is solved by a MILP solver with full constraint enforcement and certified optimality within the restricted space.

Method	Simplex iters	Branch nodes	Time (s)
MILP (no freezing)	$3.61\times 10^{6}$	$3.28\times 10^{5}$	$1149.33$
Random freezing ( $\approx 10\%$ )	$1.49\times 10^{7}$	$1.19\times 10^{6}$	$3004.75$
LLM-Assisted ( $\approx 10\%$ )	${\bm{7.07\times 10^{5}}}$	${\bm{6.26\times 10^{4}}}$	${\bm{177.60}}$

TABLE I : Comparison among plain MILP, random freezing, and LLM-Assisted.

System	Method	Avg Cost	Avg Time(s)
57-bus	MILP	$1.12\times 10^{6}$	$5.85$
57-bus	LLM	$1.12\times 10^{6}$	$4.74$
RTS	MILP	$3.30\times 10^{6}$	$1149.33$
RTS	LLM	$3.37\times 10^{6}$	$189.12$
118-bus	MILP	$1.67\times 10^{7}$	$9.06$
118-bus	LLM	$1.70\times 10^{7}$	$6.35$

TABLE II : Performance comparison of the baseline MILP and the LLM-Assisted commitment restriction without historical commitment information. Average objective value and solve time are reported over test instances.

System	Method	Avg Cost	Avg Time(s)
57-bus	MILP	$1.12\times 10^{6}$	$5.85$
	LLM	$1.12\times 10^{6}$	$3.44$
	KNN	$1.12\times 10^{6}$	$3.91$
	K-means	$1.12\times 10^{6}$	$3.84$
RTS	MILP	$3.30\times 10^{6}$	$1149.33$
	LLM	$3.30\times 10^{6}$	$171.59$

TABLE III : Average objective value and solve time of the compared UC solution strategies across different system sizes.

Parameters	MILP	K-means	KNN	LLM
Cost relative error (%)	$0.00$	$0.36$	$<0.30$	$<0.30$
Solve time (s)	$1149.33$	$714.54$	$558.34$	$171.58$
Time reduction (%)	$0.00$	$37.83$	$51.42$	$85.11$
Constraint reduction (%)	$0.00$	$33.08$	$24.46$	$36.73$
Variable reduction (%)	$0.00$	$6.4$	$12.40$	$22.52$
Node reduction (%)	$0.00$	$22.50$	$37.23$	$66.41$

TABLE IV : Comparison of different acceleration methods on the IEEE 73 bus system.

	Parameters	$10\%$	$20\%$	$30\%$	$40\%$
Ave. Time	Original MILP (s)	$1149.33$	$803.72$	$1517.87$	$1012.36$
	LLM-Assisted (s)	$171.59$	$19.34$	$80.21$	$62.85$
	Reduced Rate (%)	$85.11$	$97.59$	$94.71$	$93.79$
Max. Time	Original MILP (s)	$2678.98$	$853.44$	$2148.87$	$1337.36$
	LLM-Assisted (s)	$228.27$	$27.17$	$138.78$	$98.06$
	Reduced Rate (%)	$91.48$	$96.82$	$93.54$	$92.66$

TABLE V : Time-related results of MILP and LLM-Assisted under different load noise levels on the IEEE 73-bus system.

Parameters	$10\%$	$20\%$	$30\%$	$40\%$
Original MILP (10 5 $)	$33.01$	$32.47$	$34.17$	$34.72$
LLM-Assisted (10 5 $)	$33.01$	$33.44$	$35.47$	$35.89$
Relative Error (%)	$<0.30$	$2.99$	$3.80$	$3.37$

TABLE VI : Cost-related results of MILP and LLM-Assisted under different load noise levels on the IEEE 73 bus system.

Panel of Figure 8 — Fig. 2 : Comparison of 24-hour unit commitment schedules from the MILP baseline and the LLM-Assisted method. Rows correspond to generating units and columns to hours; white and black denote on and off states, respectively.

Fig. 3 : Solve-time comparison over 365 consecutive daily unit commitment instances. Each point corresponds to one 24-hour UC problem solved by the MILP baseline and the LLM-Assisted restriction method.

Power Systems

Power systems research studies the planning, operation, control, and protection of electrical grids and their integration with energy resources.

See all 231 papers

Apr 23, 2026

A Multi-Stage Warm-Start Deep Learning Framework for Unit Commitment

Maintaining instantaneous balance between electricity supply and demand is critical for reliability and grid instability. System operators achieve this through solving the task of Unit Commitment (UC),ca high dimensional large-scale Mixed-integer Linear Programming (MILP) problem that is strictly and heavily governed by the grid physical constraints. As grid integrate variable renewable sources, and new technologies such as long duration storage in the grid, UC must be optimally solved for multi-day horizons and potentially with greater frequency. Therefore, traditional MILP solvers increasingly struggle to compute solutions within these tightening operational time limits. To bypass these computational bottlenecks, this paper proposes a novel framework utilizing a transformer-based architecture to predict generator commitment schedules over a 72-hour horizon. Also, because raw predictions in highly dimensional spaces often yield physically infeasible results, the pipeline integrates the self-attention network with deterministic post-processing heuristics that systematically enforce minimum up/down times and minimize excess capacity. Finally, these refined predictions are utilized as a warm start for a downstream MILP solver, while employing a confidence-based variable fixation strategy to drastically reduce the combinatorial search space. Validated on a single-bus test system, the complete multi-stage pipeline achieves 100% feasibility and significantly accelerates computation times. Notably, in approximately 20% of test instances, the proposed model reached a feasible operational schedule with a lower overall system cost than relying solely on the solver.

Muhy Eddin Za'ter, Anna Van Boven, Bri-Mathias Hodge +1

University of Colorado Boulder Boulder, CO, USA

Mixed-Integer Linear Programming Power Systems

Ask PDF

Oct 18, 2026

Reinforcement learning on the EPRI AI-ccelerating Unit Commitment benchmark: diagnosing over-commitment and guiding MILP with learned policies

Reinforcement learning (RL) is increasingly proposed as a fast solver for unit commitment (UC), but standalone agents remain far more expensive than mixed-integer linear programming (MILP). In the first study of RL on the 51-generator, 72-hour benchmark of the Electric Power Research Institute (EPRI) AI-ccelerating Unit Commitment competition, we trace this gap to the agents' own exploration, close most of it, and show that the resulting policy can accelerate an exact solver. In every configuration tested, Proximal Policy Optimisation (PPO) and an action-branching Deep Q-Network (DQN) trained from scratch committed about twice the units of a thermal-only MILP optimum, at 39-64% higher cost, whereas a merit-order rule came within 1.6-7% (10-24% when unserved energy is priced). When each unit's decision is sampled independently, spare units insure against random de-commitments. Consistent with this, injecting exploration noise during fine-tuning enlarged a lean policy's fleet, a reward without shaping terms did not remove the over-commitment, and under their training noise the agents outperformed the rule subjected to comparable noise. Imitating the MILP solutions and fine-tuning conservatively brought PPO within 2.6% of the optimum, and these policies had the lowest penalised cost of any evaluated (about 5%). Fixing the fine-tuned policy's confident decisions made the MILP 3.8-8.5 times faster within 0.10-0.25% of the best solution, with a smaller worst case than an imitation-only policy fixing as many variables (1.1% against 8.9%). Because dispatch method and benchmark scope alone shift the reported optimality ratio by up to 3.6% and 9%, we propose a reporting protocol.

Maharani Devira Pramita, Aidan O'Sullivan

UCL Energy Institute, Bartlett School of Environment, Energy and Resources, University College London, Central House, 14 Upper Woburn Place, London WC1H 0NN, United Kingdom

Learning to Optimize RL for Combinatorial Optimization Power Systems Proximal Policy Optimization

Ask PDF

Sep 24, 2026

AT-SKM-Net: An Accelerated Trainable Sampling Kaczmarz-Motzkin Framework for Linear Hard-Constraint Feasibility on Dynamic Graphs

Graph-structured optimization with linear constraints is fundamental to critical infrastructure but faces scalability limits due to massive strict hard constraints and high dimensionality. While recent projection-based methods such as Trainable Sampling Kaczmarz-Motzkin Net (T-SKM-Net) guarantee feasibility, they face high computational costs in dynamic environments by processing the entire constraint set and requiring expensive matrix factorizations. To bridge this gap, we propose the Accelerated Trainable-SKM (AT-SKM) Net framework. To concentrate computation on the active constraints and eliminate redundant calculations, we introduce a hybrid sampling strategy guided by a topology-aware heterogeneous GNN model. To efficiently handle topological shifts in graph-based constraints, we employ a Cholesky Update mechanism that theoretically reduces the equality projection complexity from O(N^3) to O(N^2) under low-rank perturbations. Experiments on random geometric graphs, N-1 Security-Constrained DC-OPF, and minimum-cost gas transport problem demonstrate that AT-SKM reduces iteration counts by up to 85% and achieves 2.95x-7.29x SKM layer speedups, while maintaining zero constraint violations.

Xiaochen Zhang, Haoyu Zhu, Yao Zhang +1

Zhejiang University

Constrained Optimization

Ask PDF

Abstract

Power Systems

Explore similar work

A Multi-Stage Warm-Start Deep Learning Framework for Unit Commitment

Reinforcement learning on the EPRI AI-ccelerating Unit Commitment benchmark: diagnosing over-commitment and guiding MILP with learned policies

AT-SKM-Net: An Accelerated Trainable Sampling Kaczmarz-Motzkin Framework for Linear Hard-Constraint Feasibility on Dynamic Graphs