The rapid growth of AI/ML data centers has led to higher energy consumption and carbon emissions. The shift to renewable energy and growing data center energy demands can destabilize the power grid. Power grids rely on frequency regulation reserves, typically fossil-fueled power plants, to stabilize and balance the supply and demand of electricity. This paper sheds light on the hidden carbon emissions of frequency regulation service. Our work explores how modern GPU data centers can coordinate with power grids to reduce the need for fossil-fueled frequency regulation reserves. We first introduce a novel metric, Exogenous Carbon, to quantify grid-side carbon emission reductions resulting from data center participation in regulation service. We additionally introduce EcoCenter, a framework to maximize the amount of frequency regulation provision that GPU data centers can provide, and thus, reduce the amount of frequency regulation reserves necessary. We demonstrate that data center participation in frequency regulation can result in Exogenous carbon savings that can outweigh operational carbon emissions.

The Electric Vehicle Routing Problem with Time Windows (EVRP-TW) poses significant challenges for sustainable logistics due to its tight coupling of spatial, temporal, and energy constraints. Classical optimization methods face trade-offs: exact solvers like CPLEX ensure optimality but require prohibitive runtimes, while metaheuristics like Variable Neighborhood Search struggle with feasibility under complex constraints. We propose TERRAN, a transformer-based reinforcement learning framework for real-time and scalable EVRP-TW optimization. TERRAN integrates three key components: (1)Future-Feasibility Pruning (FFP), which proactively eliminates energy-infeasible actions by verifying reachability to charging stations or depots before each move; (2)Staged Reward Scheduling, which progressively transitions from dense auxiliary signals to task-aligned rewards to guide the agent from achieving feasibility to minimizing cost; and (3)An End-to-End Transformer-Based RL Agent Tailored for EVRP-TW, which directly integrates EV-specific constraints—including battery consumption, charging decisions, and delivery time windows—into the policy network and decoding process, enabling unified, post-processing-free optimization across varying instance scales. Experiments on Solomon benchmark instances with 5–100 customers demonstrate that TERRAN achieves 100% feasibility across all problem scales. It matches CPLEX’s optimality on 5–customer instances, achieves up to 170,000× speedups with solutions within 1.5% of optimal on 15–customer instances (0.02 s vs. 3,500 s), and delivers feasible solutions for 100–customer instances in 0.47 s, where CPLEX fails on over 80% of cases within 1 hour. These results establish TERRAN as a practical and scalable solution for real-time electric vehicle routing in complex and constraint-rich environments.

The widespread deployment of phasor measurement units (PMUs) in power transmission systems has accelerated the development of deep learning–based real-time monitoring solutions, such as event classification. Despite these advancements, recent research indicates that adversarial attacks pose a significant threat, as even minor perturbations in the input data can deceive well-trained models. In domains like computer vision, adversarial samples have been shown to transfer across different architectures, thus enabling black-box attacks via surrogate models. However, this transferability phenomenon remains largely unexplored in power system applications. In this work, we conduct a comprehensive study of adversarial transferability for power system event classification using machine learning models and a large-scale dataset. Drawing on the insights from this investigation, we propose a novel ensemble-based black-box adversarial attack that exploits transferability to achieve higher success rates and greater query efficiency. Furthermore, beyond using the Euclidean norm to measure perturbations, we incorporate signal-to-noise ratio (SNR) and maximum mean discrepancy (MMD) to enhance the robustness and depth of our perturbation analysis. Extensive experiments on a large-scale, real-world PMU dataset and state-of-the-art event classifiers highlight the effectiveness of our proposed approach.

Battery swapping stations (BSS) provide operational flexibility that can help reduce the environmental impact of electric vehicle (EV) charging. This paper presents a carbon-aware strategy that minimizes BSS emissions by scheduling battery charging based on predicted grid carbon intensity (CI) and EV swap demand. We employ online update with transformer models for CI and demand forecasting and a rolling horizon mixed integer linear programming (MILP) model for scheduling optimization. Simulations using real-world data demonstrate that the average carbon emission reductions compared to a baseline immediate-charging approach ranges from 1.8% (for the 5-slot cases) to 7.5% (for the 20-slot cases). These findings highlight the potential of predictive, carbon-aware control strategies to improve the sustainability of BSS.

Distribution grid with electric vehicle (EV) charging infrastructure can be modeled as a coupled network consisting of the cyber network, distribution grid and traffic network. The interconnection of the coupled network allows attackers to launch cyber attacks and control numerous EVs to cause severe load fluctuations, thereby affecting the normal operation of the distribution grid and the traffic flow. To evaluate the cyber-physical security of the coupled network, we present a security assessment model formulated as a coupled Discrete Event System Specification (DEVS). In this model, the evolution of the coupled network is represented as state transitions triggered by events in a discrete-time process, while the interaction is achieved through event transmission, reception and processing. To determine the redistribution of states influenced by the selection of EV charging stations and moving paths, we propose a spatial-temporal evolution mechanism. Based on the assessment model, we propose an event-triggered simulation method. The efficiency of the proposed simulation method is evaluated by comparing it with the multi-layer synchronous simulation method. Compared with the existing security assessment models, the simulation results of our model are more accurate.

Fieldbus is widely used for real-time distributed control in Industrial Control Systems (ICSs) due to its simplicity and stability. The real-world fieldbus network contains hundreds of interconnected devices, presenting a widespread network layout. Attackers can attach external intrusion devices to these communication lines to launch various attacks. In this paper, we model the fieldbus network’s channel fingerprint based on the signal’s amplitude and propose a detection method to identify potential attackers (silent intrusion devices that are eavesdropping) via channel fingerprint differences. Leveraging the passive fail-safe biasing voltage in the fieldbus network such as RS485, we can still detect the intrusion device when the fieldbus is idle (i.e., no devices are transmitting commands), which can significantly reduce the detection delay with lower sampling costs. Moreover, our method can adapt to environmental changes with little computational overhead by generating dynamic thresholds. Using a monitoring unit with stored channel fingerprints, our method can be easily deployed in fieldbus networks without occupying communication resources. The effectiveness and robustness of the proposed method have been demonstrated via extensive experiments on two real-world scenarios and one simulation scenario, where we can achieve 100% accuracy and 0% false alarm rates against various intrusion devices. Note to Practitioners—This paper is motivated by a practical need for detecting unauthorized intrusion devices in fieldbus networks. Existing detection methods face several challenges: active detection methods based on traffic analysis may disrupt normal bus communication, and it is hard to identify silent intrusion devices that are eavesdropping. Moreover, adapting these methods to changing environments is still challenging and costly. To address these issues, we leverage the inevitable amplitude differences in fail-safe biasing voltage signals and benign devices’ communication voltage signals to detect intrusion devices passively. Furthermore, to adapt to rapidly changing environments, we generate the detection thresholds dynamically based on the hypothesis testing theory. Extensive physical and simulation experiments demonstrate that the detection method against physical intrusion attacks is accurate and robust.

Adoption of renewable energy in power grids introduces stability challenges in regulating the operation frequency of the electricity grid. Thus, electrical grid operators call for provisioning of frequency regulation services from end-user customers, such as data centers, to help balance the power grid’s stability by dynamically adjusting their energy consumption based on the power grid’s need. As renewable energy adoption grows, the average reward price of frequency regulation services has become much higher than that of the electricity cost. Therefore, there is a great cost incentive for data centers to provide frequency regulation service. Many existing techniques modulating data center power result in significant performance slowdown or provide a low amount of frequency regulation provision. We present PowerMorph , a tight QoS-aware data center power-reshaping framework, which enables commodity servers to provide practical frequency regulation service. The key behind PowerMorph is using “complementary workload” as an additional knob to modulate server power, which provides high provision capacity while satisfying tight QoS constraints of latency-critical workloads. We achieve up to 58% improvement to TCO under common conditions, and in certain cases can even completely eliminate the data center electricity bill and provide a net profit.

In this paper, we present and illustrate some new tools for rigorously analyzing training data selection methods. These tools focus on the information theoretic losses that occur when sampling data. We use this framework to prove that two methods, Facility Location Selection and Transductive Experimental Design, reduce these losses. These are meant to act as generalizable theoretical examples of applying the field of Information Theoretic Deep Learning Theory to the fields of data selection and active learning. Both analyses yield insight into their respective methods and increase their interpretability. In the case of Transductive Experimental Design, the provided analysis greatly increases the method’s scope as well.

Providing ride-hailing services with electric vehicles can help reduce greenhouse gas emissions and solve the last mile problem. This paper develops a reinforcement learning based algorithm to operate a community owned electric vehicle fleet, which provides ride-hailing services to local residents. The goals of operating the electric vehicle fleet are to minimize customer waiting time, electricity cost, and operational costs of the vehicles. A novel framework characterized by decentralized learning and centralized decision making is proposed to solve the electric vehicle fleet dispatch problem. The decentralized learning process allows the individual vehicles to share their operating experiences and deep neural network model for state-value function estimation, which mitigates the curse of dimensionality of state and action domains. The centralized decision making framework converts the vehicle fleet coordination problem into a linear assignment problem, which has polynomial time complexity. Numerical study results show that the proposed approach outperforms the benchmark algorithms in terms of societal cost reduction.

This article considers the subject of information losses arising from the finite data sets used in the training of neural classifiers. It proves a relationship between such losses as the product of the expected total variation of the estimated neural model with the information about the feature space contained in the hidden representation of that model. It then bounds this expected total variation as a function of the size of randomly sampled data sets in a fairly general setting, and without bringing in any additional dependence on model complexity. It ultimately obtains bounds on information losses that are less sensitive to input compression and in general much smaller than existing bounds. This article then uses these bounds to explain some recent experimental findings of information compression in neural networks that cannot be explained by previous work. Finally, this article shows that not only are these bounds much smaller than existing ones, but they also correspond well with experiments.

Electricity theft is the third largest form of theft in the United States. It not only leads to significant revenue losses, but also creates the risk of fires and fatal electrical shocks. In the past, utilities have fought electricity theft by sending field operation groups to conduct physical inspections of electrical equipment based on suspicious activity reported by the public. However, the recent rapid penetration of advanced metering infrastructure makes it possible to detect electricity theft by analyzing the information gathered from smart meters. In this paper, we develop a physically inspired data driven model to detect electricity theft with smart meter data. The main advantage of the proposed model is that it only leverages the electricity usage and voltage data from smart meters instead of unreliable parameter and topology information of the secondary network. Hence, a speedy and widespread adoption of the proposed model is feasible. We show that a modified linear regression model accurately captures the physical relationship between electricity usage and voltage magnitude on the Kron-reduced distribution secondaries. Our results show that electricity theft on a distribution secondary will lead to negative and positive residuals from the regression for dishonest and honest customers, respectively. The proposed model is validated with real-world smart-meter data. The results show that the model is effective in identifying electricity theft cases.

In this paper, an improved multiperiod risk-limiting dispatch (IMRLD) is proposed as an operational method in power systems with high percentage renewables integration. The basic risk-limiting dispatch (BRLD) is chosen as an operational paradigm to address the uncertainty of renewables in this paper due to its three good features. In this paper, the BRLD is extended to the IMRLD so that it satisfies the fundamental operational requirements in the power industry. In order to solve the IMRLD problem, the convexity of the IMRLD is verified. A theorem is stated and proved to transform the IMRLD into a piece-wise linear optimization problem that can be efficiently solved. In addition, the locational marginal price of the IMRLD is derived to analyze the effect of renewables integration on the marginal operational cost. Finally, two numerical tests are conducted to validate the IMRLD.

Accurate network and phase connectivity models are crucial to distribution system analytics, operations and planning. Although network connectivity information is mostly reliable, phase connectivity data is typically missing or erroneous. In this paper, an innovative phase identification algorithm is developed by clustering of voltage time series gathered from smart meters. The feature-based clustering approach is adopted where principal component analysis is first carried out to extract feature vectors from the raw time series. A constrained k-means clustering algorithm is then executed to separate customers/smart meters into various phase connectivity groups. The algorithm is applied on a real distribution feeder in Southern California Edison's service territory. The accuracy of the proposed algorithm is over 90%.

Buildings account for nearly 40 percent of the total energy consumption in the United States. As a critical step toward smart cities, it is essential to intelligently manage and coordinate the building operations to improve the efficiency and reliability of overall energy system. With the advent of smart meters and two-way communication systems, various energy consumptions from smart buildings can now be coordinated across the smart grid together with other energy loads and power plants. In this paper, we propose a comprehensive framework to integrate the operations of smart buildings into the energy scheduling of bulk power system through proactive building demand participation. This new scheme enables buildings to proactively express and communicate their energy consumption preferences to smart grid operators rather than passively receive and react to market signals and instructions such as time varying electricity prices. The proposed scheme is implemented in a simulation environment. The experiment results show that the proactive demand response scheme can achieve up to 10 percent system generation cost reduction and 20 percent building operation cost reduction compared with passive demand response scheme. The results also demonstrate that the system cost savings increase significantly with more flexible load installed and higher percentage of proactive customers participation level in the power network.