CS Paper Radar

Decision-Focused Learning via Neural Optimization Layers for Battery-Assisted Grid-Import Smoothing in AI Data Centers

Xinda Zheng, Canchen Jiang, Frits de Nijs, Edward Lam, Markus Wagner

eess.SY cs.LGmath.OC ·submitted 2026-11-10 07:41 UTC · v1

Abstract

Rapid AI data-center power fluctuations challenge grid integration. Sustained grid-import smoothing requires allocating limited battery power and energy under uncertain demand. We formulate risk-aware BESS scheduling that coordinates advance reserve commitments and demand-responsive dispatch, accounting for operating costs, grid-ramp penalties, and their conditional value-at-risk. We develop battery scheduling decision-focused learning (BS-DFL) with a constraint-aware neural optimization layer trained directly from this objective without optimal-solution labels. The trained layer is then frozen to guide demand-predictor learning through scheduling costs and ramping risk, enabling solver-free rolling-horizon scheduling. In a 10,000-minute simulation using loads derived from measured AI workloads, BS-DFL reduces mean absolute minute-to-minute grid-import changes by 71.70% and total scheduling cost by 18.01% relative to an MSE-trained predictor coupled with a mixed-integer linear programming solver, with an average decision time of 18.77 ms.

Closest papers

Nearest neighbours by meaning across the whole index.

Decision-Focused Learning for Optimal Battery Scheduling Under Load and Photovoltaic Power Uncertainty in Microgrids

eess.SY ·2026-10-10 ·Marios Saleptsis, Antoine Lesage-Landry, Marco Mussetta, Sonia Leva
similarity 0.75
Abstract · 20 pages, 13 figures, submitted to Sustainable Energy, Grids and Networks

The increasing penetration of renewable energy sources in microgrids demands accurate day-ahead forecasting to support optimal battery energy storage system (BESS) scheduling. Yet the conventional predict-then-optimize paradigm trains forecasters on statistical accuracy metrics that are agnostic to the asymmetric and time-varying cost impacts of forecast errors. This work proposes a decision-focused learning (DFL) framework that trains gated recurrent unit forecasters for both load demand and photovoltaic generation directly on the downstream operational cost of a two-stage energy management system, where day-ahead BESS commitments are fixed prior to real-time balancing. Gradients are backpropagated from the real-time operational cost through both optimization stages to the forecaster weights via implicit differentiation, enabling the forecasters to internalize the penalty structure of the scheduling problem. The framework is validated on the MG2Lab experimental microgrid testbed at Politecnico di Milano across two operational scenarios of increasing grid constraint and eight forecasting configurations comprising single-source and combined-forecaster settings. Results illustrate that the DFL-trained pipeline reduces operational cost by 3.78% under normal grid conditions and by 30.60% under grid-constrained conditions relative to the accuracy- optimized baseline, while nearly eliminating load shedding events. Notably, the DFL forecasters exhibit higher statistical errors than their accuracy-trained counterparts yet achieve consistently lower operational costs, confirming that forecast accuracy and decision quality do not share a monotonic relationship in constrained microgrid scheduling.

0.12

Comparative Evaluation of MILP, MPC, and Reinforcement Learning for Commercial Battery Dispatch Under Time-of-Use Tariffs

eess.SY ·2026-09-13 ·Hafiz Majid Hussain, Wajiha Samar, Lurian Klein, Pedro Nardelli
benchmark similarity 0.71
Abstract

Battery energy storage systems (BESS) paired with rooftop photovoltaics (PVs) can deliver measurable cost savings under time-of-use (TOU) electricity tariffs; however, the relative performance of model-based and model-free dispatch strategies remains insufficiently benchmarked on full-year, real-world commercial datasets. This paper presents a full-year (2023) comparative evaluation of three BESS dispatch approaches using data from a commercial PV installation operating under a TOU tariff. The examined strategies include: (i) a mixed-integer linear programming (MILP) formulation with perfect foresight, providing an oracle performance benchmark under the assumed model; (ii) a model predictive control (MPC) scheme based on a day-ahead persistence forecast, representing a low-complexity deployable approach; and (iii) a soft actor-critic (SAC) deep reinforcement learning agent trained under causal information constraints. The MILP benchmark achieves an annual cost reduction of 24.6\% relative to a no-storage baseline. The persistence-based MPC approach recovers 99.2\% of this benchmark using only prior-day data. In contrast, the evaluated SAC agent yields an annual cost higher than the no-storage baseline. This outcome is analyzed in the context of known challenges in reinforcement learning for energy systems, including limited observability and reward design. Overall, the results indicate that, for the studied dataset and tariff structure, persistence-based MPC captures nearly all achievable economic benefits under practical deployment constraints, whereas the considered RL configuration does not yield competitive performance under the same information limitations.

0.21

Frequency-Calibrated Decision-Focused Learning for BESS Dispatch Under Renewable Uncertainty

eess.SY ·2026-10-30 ·Zihan Meng, Chenxi Hu, Xin Cui, Francis Yunhe Hou
hot direction similarity 0.71
Abstract

Battery energy storage system (BESS) flexibility depends on state of charge and intertemporal operating constraints. Renewable forecast errors of similar magnitudes can lead to different BESS dispatch results. An error may alter early charging or discharging decisions and reduce the flexibility available for subsequent regulation, while another may have little effect on the resulting schedule. To account for these unequal consequences, this paper proposes Frequency-Calibrated Decision-Focused Learning (FC-DFL), which trains renewable predictors using the realized operating costs of their induced BESS schedules. Gaussian smoothing and solver-based zeroth-order estimation provide training feedback without requiring differentiating through the non-differentiable mixed-integer BESS dispatch problem. The same decision feedback is further combined with prediction sensitivity to develop a decision-oriented XAI method that identifies the input features and forecast periods most consequential to BESS operation. Case studies using German wind and solar data show that FC-DFL reduces the integrated absolute frequency-deviation metric by 6.05% and decision regret by 23.92% relative to accuracy-based predict-then-optimize method. The XAI analysis further demonstrates that the identified high-impact forecast periods produce 2.84 times larger average absolute changes in realized operating cost than low-impact periods. These findings highlight the importance of incorporating downstream decision-making outcomes into renewable forecasting and demonstrate the value of FC-DFL in improving BESS regulation and providing operationally meaningful forecast explanations.

0.24

Electric Vehicle Scheduling and Carbon-Aware Energy Dispatch for Charging Stations

eess.SY ·2026-10-27 ·Fulong Yao, Yiming Xu, Maurizio Albano, Liana Cipcigan, Naeima Hamed +2 more
similarity 0.69
Abstract

Electric vehicle (EV) charging stations integrating photovoltaic (PV) generation, battery storage, and bidirectional grid exchange must coordinate access to limited chargers with energy supply while considering both energy cost and carbon emissions. This paper proposes a solver-free greedy heuristic (SFGH) for dynamic vehicle scheduling (VS) and a twin delayed deep deterministic policy gradient (TD3) controller for carbon-aware energy dispatch (ED). SFGH assigns EVs according to reductions in service delays, and TD3 controls battery operation to balance energy costs and carbon emissions while meeting the resulting charging demand. The model allows both PV and battery exports, accounting for export revenue and carbon credits. Simulations on year-long case studies show that SFGH--TD3 achieves the lowest accumulated service delay, energy cost, and carbon emissions among the evaluated heuristic and multi-agent deep reinforcement learning (DRL) configurations.

0.24

Mitigating the Grid Impact of Gigawatt-Scale AI Data Centers via High-Fidelity Model Flexibility

eess.SY math.OC ·2026-11-18 ·Eihab E. E. Ahmed, Saeed Manshadi
hot direction similarity 0.68
Abstract

Gigawatt-scale artificial intelligence data centers (AIDCs) are being connected at a single point of interconnection (POI) faster than transmission can be reinforced, yet studies of their flexibility rely on DC power flow and simplified load envelopes that omit voltage, cooling dynamics, and workload deadlines. This paper addresses this gap by embedding a high-fidelity AIDC model, with separate inference and training pools, hall thermal mass, PV, a battery energy storage system (BESS) with reactive capability, and a STATCOM, in a five-minute multi-period AC optimal power flow. Simulations on the RTS-GMLC test system show that active-power flexibility mitigates local price spikes and relieves congestion, whereas reactive support from the BESS inverter or a STATCOM is needed to keep the POI voltage away from sagging around its 0.95~pu limit. Sensitivity analyses show that cost-driven charging of large batteries can induce new morning congestion and that storage largely substitutes for wider workload deferral.

0.30

End-to-End Battery Dispatch with Exact Rainflow Degradation via Mixed-Integer Differentiable Predictive Control

eess.SY ·2026-09-11 ·Eshagh Safarzadeh Ravajiri, Jan Drgona, Mahdi Mehrtash, Benjamin F. Hobbs
top venue similarity 0.68
Abstract · 12 pages. Submitted to IEEE Transactions on Smart Grid

Optimal dispatch of battery energy storage systems requires balancing energy arbitrage against cycle-induced degradation, which is accurately quantified through rainflow cycle counting. However, rainflow's combinatorial, nondifferentiable algorithm is incompatible with both convex optimization and gradient-based neural network training. We present a self-supervised mixed-integer differentiable predictive control framework that trains neural policies directly on exact rainflow degradation through a novel differentiable rainflow layer combining exact gradients at state-of-charge extrema with dense proxy gradients on incremental changes, enabling stable end-to-end training while preserving true degradation physics. A mixed-integer differentiable architecture enforces power balance, dynamics, and mode exclusivity, with a safety filter guaranteeing constraint satisfaction. We evaluate the framework on 3,650 real battery-day scenarios (a 10-battery fleet over 365 days) spanning three utility regions (SDG&E California, Xcel Energy Colorado, APS Arizona) with diverse time-of-use pricing structures. A single-battery trained policy achieves a 0.35% performance gap on its training distribution with over 200x speedup, while fleet-wide training generalizes across all households and utility regions, achieving a 3.33% performance gap with 564x computational speedup (62 seconds vs. 9.7 hours) and 100% feasibility. The millisecond-scale inference reduces computation by over two orders of magnitude compared to mixed-integer solvers, enabling practical deployment at scale.

0.71

Carbon-Aware Electric Vehicle Routing Problem with Decision-Focused Learning

cs.LG math.OC ·2026-10-29 ·Jaume Reixach, Wanru Chen, Jiaming Wu, Balázs Kulcsár
similarity 0.68
Abstract

Electric vehicles produce no tailpipe emissions, but charging them produces emissions that depend on the time-varying generation mix of the electricity consumed. We study an Electric Vehicle Routing Problem with a carbon-aware objective that accounts for charging-related emissions alongside conventional routing costs. The problem determines routing, charging, and waiting decisions, allowing charging demand to be shifted toward periods with lower grid carbon intensity. Because these decisions depend on forecasts of future carbon intensity, we propose a decision-focused learning approach so that the training of the model employed for predicting the carbon intensity of the grid over time is solver-aware. In particular, we train a linear predictor using approximate SPO+ directions and employ a time-limited Large Neighborhood Search followed by exact postprocessing as the underlying solver. Experiments on realistic Germany-based instances show that explicitly considering charging-related emissions reduces realized emissions by 28.76% on average relative to carbon-blind optimization, although at the expense of higher non-carbon operating cost. Moreover, decision-focused training reduces realized total cost by 2.30% on average relative to conventional prediction-focused learning and performs better on 75.5% of the test instances, despite substantially lower pointwise predictive accuracy. Finally, we show that its average advantage remains positive across varied solver configurations and carbon-cost weights.

0.26

Speed-to-Power Dominates Energy Costs for AI Data Centers: Technoeconomic Analysis of Powering AI in the Continental United States

cs.AI eess.SY ·2026-11-12 ·Alex C. Newkirk, Daniel L. Gerber, Erica R. H. Fuchs, Alex Hubbard, M. Granger Morgan +3 more
code/datahot direction similarity 0.68
Abstract · 80 pages incl. SI

The rapid growth of artificial intelligence is driving demand for new data centers at large scale and on short deployment timelines. To power them, firms connect to the commercial grid or contract generation directly, but interconnection and gas-turbine delays now incentivize on-site generation. We compare grid-connected and off-grid data center energy infrastructure (AC- and DC-coupled solar-plus-storage, islanded natural gas, and grid interconnection) across the continental U.S. Using an open-source technoeconomic framework built on empirically calibrated AI training loads, climate-driven cooling, stage-by-stage power conversion, and physics-based battery degradation, we model designs providing equivalent energy services. Our central contribution is a speed-adjusted cost of energy that values the compute unlocked by faster construction, making options with different deployment timelines directly comparable on one physically consistent basis. We find DC coupling reduces lifetime energy costs by 18% ($0.09/ kWh) on average versus AC counterparts by eliminating inverters and improving efficiency. Though rarely cheapest on unadjusted energy cost, DC-coupled microgrids' faster deployment makes them the lowest-cost modeled option in 85% of U.S. locations once foregone compute is valued, or 55% under accelerated natural gas timelines. For AI infrastructure, deployment speed rather than unit energy cost determines the economically preferred path to power.

0.31