Backend
Profile Screening and Frequency-Secure Operating Limit Derivation for Data Center Microgrids
arXiv cs.SY0 views
arXiv:2611.23504v1 Announce Type: new Abstract: Rapid power variations from artificial intelligence (AI) workloads pose frequency-security challenges for islanded data-center microgrids. This paper proposes a generator step- response-based framework to rapidly screen candidate load pro- files and derive frequency-secure operating limits for IT workload scheduling. Identified generator responses are superimposed to assess the cumulative frequency effects of successive load changes without repeated dynamic simulations. The framework identifies potentially unsafe intervals and determines allowable load magni- tudes, ramp rates, and minimum event spacing, while accounting for prediction uncertainty and governor limits. Validation on a 2.5 MVA islanded microgrid with eight generator configurations and 14 synthetic AI workload profiles shows that 94.75% of 40,208 assessment windows receive decisive screening labels, with no frequency-violating windows classified as secure.
Source:https://arxiv.org/abs/2611.23504PreviousAdaptive Line Switching for Cascade Mitigation in Power Systems via Graph Reinforcement LearningNextReinforcement Learning Optimized Flight Control for Interception of High-Altitude Maneuvering Targets
Related posts
Learning to Decide with AI Assistance under Human-Alignment
0arXiv cs.HC
From Defense to Advocacy: Empowering Users to Leverage the Blind Spot of AI Inference
0arXiv cs.HC
Noise-induced enhancement of regime lifetimes -- A data-driven approach using deterministic trajectories
0arXiv cs.NA
fOGA: An Orthogonal Greedy Algorithm for Fractional Laplacian Problems
0arXiv cs.NA
Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts
0arXiv stat.ML
Kinetic Langevin Splitting Schemes for Constrained Sampling
0arXiv stat.ML