Energy Informatics最新文献

{"title":"Causal analysis of speed effects on energy consumption and driving range in electric cargo vehicles","authors":"Mrugank Gandhi,&nbsp;Archana Y. Chaudhari,&nbsp;Rahesha Mulla","doi":"10.1186/s42162-026-00667-0","DOIUrl":"10.1186/s42162-026-00667-0","url":null,"abstract":"<div><p>Accurate estimation of energy consumption for EVs and the remaining driving range is crucial for effective energy management. Most existing studies on EV energy consumption estimation are based on correlation-based prediction. However, they do not offer much insight into the causal effects of the driving factors. This study presents a causal machine learning framework that considers the vehicle speed as an intervention to estimate the effect on the energy consumption and the remaining driving range of the EV. Telemetry data obtained from a Musoshi Pop-Up Mini electric cargo vehicle at different speed regimes, payload, and environmental conditions are used to evaluate the proposed framework. This involves the construction of a domain-informed directed acyclic graph for confounding identification and the implementation of Orthogonal Double Machine Learning via the integration of DoWhy and EconML LinearDML estimator for Average Treatment Effects (ATE) and Conditional Average Treatment Effects (CATE). Multiple specifications of the nuisance model and robustness checks, such as placebo refutation, subset stability, overlap checks, and regularization sensitivity, are carried out to ensure the validity of the obtained causal effects. The findings reveal a statistically stable causal relationship between vehicle speed and segment-level energy consumption, with significant heterogeneity observed depending on the operating conditions, where the effect of road slope was found to be the main moderator. The energy effects are further propagated to the remaining driving range to obtain the operational interpretation. Overall, the proposed framework reveals that causal intervention-aware modeling can produce reliable and interpretable insights for range-aware control and smart energy management in electric vehicles.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00667-0.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148380408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Frugal and explainable artificial intelligence for microgrid energy management systems: a bibliometric view","authors":"Serge Raoul Dzonde Naoussi,&nbsp;François Dieudonné Mengue,&nbsp;David Tsuanyo,&nbsp;Bernard Kamsu Foguem","doi":"10.1186/s42162-026-00654-5","DOIUrl":"10.1186/s42162-026-00654-5","url":null,"abstract":"<div><p>As decentralized energy systems become pivotal for global decarbonization, Microgrid Energy Management Systems (EMS) are increasingly adopting Artificial Intelligence (AI) to cope with renewable intermittency, load uncertainty, and multi-objective operational constraints. While deep learning has substantially improved predictive capabilities and control performance, it has introduced two critical externalities: the limited transparency of “black-box” models and escalating computational energy costs. This study conducts a systematic bibliometric analysis of 2,150 records from the Scopus database (2011–2025), using VOSviewer and the Bibliometrix R package to map the convergence and divergence of Explainable AI (XAI) and Frugal AI within the microgrid domain. Our findings reveal a significant dichotomy: while research into interpretability is experiencing exponential growth due to regulatory and operational trust requirements, “frugality” (computational efficiency) remains a peripheral concern, rarely prioritized in model design despite the resource constraints of edge deployment. We argue that the current literature overlooks the environmental footprint of the algorithms themselves. To the best of our knowledge, this is the first bibliometric study that jointly examines frugality and explainability in AI for microgrid energy management, rather than treating them as separate lines of inquiry. Consequently, we propose a consolidated research agenda focusing on hybrid architectures that reconcile high-performance forecasting with lightweight, transparent design, essential for the sustainable deployment of intelligent microgrids. Beyond mapping the field, the proposed agenda is intended to guide researchers and practitioners toward deployable EMS solutions that respect both hardware constraints and emerging requirements for algorithmic transparency.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00654-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148381031","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Flexible clustering of substations for accurate and rapid hybrid simulation of district heating","authors":"Dubon Rodrigue,&nbsp;Mohamed Tahar Mabrouk,&nbsp;Bastien Pasdeloup,&nbsp;Patrick Meyer,&nbsp;Bruno Lacarrière","doi":"10.1186/s42162-026-00664-3","DOIUrl":"10.1186/s42162-026-00664-3","url":null,"abstract":"<div><p>District Heating Networks (DHNs) are crucial to decarbonizing the heat supply sector, with the evolution toward 4th and 5th generation systems offering significant potential for efficiency. However, the increasing complexity of these modern systems makes high-fidelity dynamic thermo-hydraulic simulation computationally intensive, particularly for large-scale networks. These simulations are essential for key applications such as thermal loss prediction, supply temperature optimization, operational planning and during the sizing phase of the network. Recent research has utilized Machine Learning (ML)-based surrogate models to replace substation clusters, reducing spatial complexity and accelerating simulations. Yet, the efficacy of this spatial reduction is highly sensitive to the clusters definition. Poorly selected clusters degrade the surrogate models accuracy and undermine the simulation performance. This paper proposes a flexible, task-driven graph clustering methodology specifically designed for ML-based spatial reduction. We introduce physics-informed distance metrics that encode the primary drivers of ML surrogate model errors. These distance metrics are leveraged within a hierarchical agglomerative clustering framework. The methodology is evaluated across 16 DHNs with diverse topological and thermal characteristics. Our results reveal a clear, generalizable trade-off between simulation accuracy and computational time reduction, allowing the integration of user-preferences. On an independent validation DHN, a high physical accuracy preference yielded a deviation of 6.45 MWh in total thermal energy production estimation (3.76% of total thermal losses within the network), albeit with modest computational gains. Expressing the error relative to thermal losses, rather than total production, provides a more meaningful measure of the spatial reduction’s impact on the key operational objectives. Conversely, a high spatial reduction preference reduced the network from 123 to 5 remaining physical nodes and achieved a 91% decrease in computational time, at the cost of a 45.24 MWh deviation in total thermal energy production estimation, representing 26.37% error relative to thermal losses. These findings demonstrate that the proposed clustering framework provides a robust and flexible tool to calibrate the balance between physical fidelity of the simulation and computational speed across varying DHN generations.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00664-3.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148380320","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A just energy location framework for equitable and sustainable electric vehicle charging infrastructure planning: techno-economic assessment and policy benchmarking in Kerala","authors":"Shambhu Sajith,&nbsp;R. S. Aswani","doi":"10.1186/s42162-026-00666-1","DOIUrl":"10.1186/s42162-026-00666-1","url":null,"abstract":"<div><p>The lack of even distribution of charging points in the emerging economies such as India hinders EVs use. The paper suggests the Just Energy Location (JEL) framework - a new conceptual and analytical tool to combine classical Location Theory and the concepts of Energy Justice to inform the planning of equitable, efficient, and participative EV charging infrastructure. This framework is operationalised using three pillars, including spatial and technical optimisation, equitable access and inclusion, and participatory governance that have measurable indicators based on both location-theoretic and energy-justice literature. These indicators are used to test actual policy settings, which allow a systematic assessment of policy settings, as opposed to an impressionistic one. The study carries out a techno-economic analysis by using HOMER software to analyze four Hybrid Renewable Energy System (HRES) based on a standalone EV charging station in Technopark, Thiruvananthapuram, Kerala. The findings find vanadium flow batteries as an interesting long-term alternative due to their ability to provide significantly greater cost stability and resiliency to the supply chain, whereas wind-solar-lithium-ion battery is most cost-effective. The configurations are also assessed based on the alignment of each variable to the JEL framework pillar of equity and governance, as well as on LCOE and capital cost, and the study conceptual and empirical pillars are combined directly. The paper compares the policies of China, the United States and European Union concerning the EV charging policies to the policy environment in Kerala through the structured rubric in the JEL framework. The results provide direct implications to the state planners, DISCOMs, and policymakers interested in promoting sustainable, inclusive EV implementation, and are applicable to urban environments with similar energy and transport interactions, in general.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-05-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00666-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148323371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A comprehensive framework for modeling building occupancy: technologies, methods, and research challenges","authors":"Geoffrey Bourgoin,&nbsp;Kevin Micheneau,&nbsp;Abhinandana Boodi,&nbsp;Karim Beddiar","doi":"10.1186/s42162-026-00660-7","DOIUrl":"10.1186/s42162-026-00660-7","url":null,"abstract":"<div><p>Occupancy information is a key element for improving building energy performance and occupant comfort through occupancy-centric control strategies. However, the rapid development of occupancy-related research over the past years has led to fragmented terminology and heterogeneous analytical perspectives, which complicate consistent comparison and interpretation of results. To address this issue, this paper proposes an analytical occupancy information framework that structures the literature according to complementary dimensions: occupancy scale, information level, acquisition strategy, modeling approach, temporal dimension, and system constraints. Based on this framework, a systematic review of the literature is conducted under a novel analysis axis based on occupancy scale. The analysis highlights trade-offs between information richness, deployability, privacy constraints, and control relevance across scales. Occupancy presence and estimation appear as the most widely investigated and practically exploitable information levels for energy-oriented applications, particularly at zone and multi-zones scales, where they are more directly integrated into HVAC control strategies. More detailed information levels, such as localization, tracking, activity, comfort, and behavior, provide complementary insights but remain constrained by sensing complexity, privacy considerations, and challenges in generalization. Across occupancy scales, black-box modeling approaches are the most frequently adopted in the literature, while gray-box and knowledge-based approaches offer potential advantages in terms of interpretability and adaptability. Overall, the proposed framework provides a unified perspective on occupancy information modeling, clarifies the respective contributions and limitations associated with each occupancy scale, and identifies key challenges and research directions toward scalable, privacy-aware, and energy-efficient building control systems.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00660-7.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148172861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A comparative analysis of electricity demand forecasting using modelling methods, factors, and horizon: a review study of developing and developed countries","authors":"Zara Omar,&nbsp;Liwan Liyanage","doi":"10.1186/s42162-025-00596-4","DOIUrl":"10.1186/s42162-025-00596-4","url":null,"abstract":"<div><p>Electricity plays a crucial role in driving economic growth and ensuring the stability of a country. While electricity demand and supply tend to remain stable in developed nations, many developing countries continue to face persistent electricity crises. The primary objective of this meta-analysis is to identify gaps in the literature related to electricity demand forecasting methodologies and to examine the recommendations proposed by researchers in the context of three selected countries. This study presents a comprehensive overview of statistical and machine learning models used in electricity demand forecasting, based on a review of 107 articles published between 2005 and 2023, highlighting the common variables, influencing factors, and forecasting horizons considered. The literature reveals that research motivations differ according to the resources and contextual factors of each country: in developing countries, studies are often driven by economic indicators, as well as social and household parameters, and predominantly employ traditional prediction methods; in contrast, studies in developed countries incorporate climate variables alongside other indicators and tend to adopt modern modelling techniques. Findings reveal that developing countries predominantly rely on traditional statistical methods driven by economic, social, and household indicators, whereas developed countries integrate climate variables and increasingly adopt advanced modelling techniques. The study identifies methodological gaps and provides insights to enhance the accuracy and applicability of future electricity demand forecasting models.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-025-00596-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148172860","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Topology-aware graph-attentive one-class anomaly detection for physics-based cybersecurity monitoring in photovoltaic systems","authors":"Danilo Greco,&nbsp;Giovanni Battista Gaggero","doi":"10.1186/s42162-026-00661-6","DOIUrl":"10.1186/s42162-026-00661-6","url":null,"abstract":"<div><p>Photovoltaic (PV) plants increasingly operate as software-defined cyber-physical systems, where firmware updates, remote supervision, and digital control loops expand the attack surface. A key challenge for practical monitoring is the scarcity and diversity of labelled cyberattack traces: PV systems run predominantly under normal conditions, while faults and intrusions are rare and continually evolving. This work presents a <i>topology-aware, graph-attentive one-class</i> framework for physics-based cybersecurity monitoring in PV systems. We represent operating samples as nodes of a <i>multiscale</i> <i>k</i>-nearest-neighbour graph and learn a compact manifold of normal behaviour with a <i>Graph Attention Autoencoder</i> regularised by (i) robust reconstruction, (ii) SVDD-style latent compactness, and (iii) graph smoothness. Anomaly detection uses an ensemble score that combines reconstruction discrepancy and latent-space deviations, yielding a stable decision function across heterogeneous cyber-physical perturbations. We evaluate the approach on <i>Photo-Set</i>, a benchmark dataset for physics-based PV cybersecurity monitoring. Across the available Photo-Set test datasets, the proposed model provides strong ROC and precision–recall characteristics and competitive (often superior) F<span>(_1)</span> and MCC relative to Isolation Forest, One-Class SVM, and Local Outlier Factor. We discuss failure modes on subtle scenarios, ablations that isolate the contribution of graph construction and latent regularisation, and practical guidance for deployment in resource-constrained monitoring pipelines.</p></div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00661-6.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148008851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MDJA-Trans: a multivariate data joint analysis transformer architecture with dynamic gating for load-aware low-voltage AC Arc fault detection","authors":"Jing Li,&nbsp;Nawaraj Kumar Mahato,&nbsp;Yubin Guo,&nbsp;Junfeng Yang","doi":"10.1186/s42162-026-00663-4","DOIUrl":"10.1186/s42162-026-00663-4","url":null,"abstract":"<div>\u0000 \u0000 <p>To address the challenges of multivariate signal coupling, noise interference, and load adaptability in arc fault detection within low-voltage distribution systems, this paper introduces a novel arc fault detection architecture, termed Multivariate Data Joint Analysis with Transformer (MDJA-Trans) architecture. The architecture employs Noise-Suppressing Time Convolutional Learning (NS-TCL) to dynamically mitigate noise and extract multiscale arc features through frequency-domain masking. Load-Aware Cross-Variate Attention Fusion (CVAF) facilitates dynamic alignment of voltage and current features, while Dynamic Gating Decision (DGD) integrates electrical features with load attribute embeddings for robust arc fault classification. This approach enhances physical interpretability and load generalization. Extensive experiments demonstrate superior noise immunity and cross-load adaptability, achieving an arc fault detection accuracy and F1-Score of 98.55%. The proposed solution will provide technical support for improving household safety and reducing electrical fire risks.</p>\u0000 </div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00663-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147829913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-sensor informatics workflow for appliance-level energy use, comfort, and mixed-mode cooling in a composite-climate home","authors":"Sahil Chilana,&nbsp;Aviruch Bhatia,&nbsp;Vishal Garg,&nbsp;Jyotirmay Mathur","doi":"10.1186/s42162-026-00662-5","DOIUrl":"10.1186/s42162-026-00662-5","url":null,"abstract":"<div>\u0000 \u0000 <p>This study presents a year-long, multi-sensor residential monitoring dataset from a composite-climate home in Mohali, India, integrating appliance-level electricity use, indoor temperature–humidity measurements, and outdoor weather observations into an hourly residential energy informatics workflow. Nine appliances were monitored at 5-minute resolution and processed through a reproducible pipeline involving timestamp harmonization, anomaly handling, multi-source alignment, and feature extraction, yielding an analytics-ready dataset spanning 8,736 hourly observations (July 2024–June 2025). The monitored appliances consumed 845.25 kWh (58% of total household electricity use), with the refrigerator (21.95%) and air conditioner (cooling + heating; 19.07%) as the dominant monitored end uses. Psychrometric mapping quantified mixed-mode operation, distinguishing operating regions for Air Conditioner (AC) cooling (n=460 hours), evaporative cooling (n=172 hours), and AC heating (n=45 hours). Weather sensitivity analysis showed strong temperature dependence in baseline and seasonal loads (effect of outside temperature on refrigerator r=0.793; effect of outside temperature on geyser r=-0.694; slope -0.048 kWh/<span>(^{circ })</span>C), demonstrating the value of fusing weather variables into appliance analytics. Indoor comfort evaluation using the India Model for Adaptive Comfort (IMAC) indicated that 80.8% of summer hours and 51.7% of monsoon hours with valid measurements fell within adaptive comfort limits, with winter interpretation constrained by modeling assumptions and sensor availability. The primary contribution is a transparent, transferable workflow that converts heterogeneous residential sensing streams into a unified analytics dataset suitable for season-responsive appliance modeling, comfort-aware interpretation, and future multi-home deployments.</p>\u0000 </div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00662-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147830027","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Towards carbon-aware AI: a systematic prisma review and taxonomy of green architectures, hardware life-cycle, and energy-efficient algorithms","authors":"Raghavendra M. Devadas,&nbsp;Sowmya T","doi":"10.1186/s42162-026-00651-8","DOIUrl":"10.1186/s42162-026-00651-8","url":null,"abstract":"<div>\u0000 \u0000 <p>The increasing computational requirements in modern Artificial Intelligence (AI) development have raised the stakes when it comes to the environmental sustainability of machine learning use and application. The topic that has received relatively little treatment in previous research is the carbon footprint of AI systems, but it is not studied continuously across algorithms, hardware, life cycle, and use. This work takes stock of the state-of-the-art to provide a comprehensive review of carbon-aware AI from the bottom up for the entire gamut of computation. Consistent with PRISMA principles, by way of systematic search across prominent academic databases and repositories (2018–2025), we identified 784 unique citations and extracted 62 studies that satisfied pre-established inclusion criteria. These studies were organized by four domains: algorithms for energy (20), hardware and accelerators (15), Life-cycle assessment (LCA) (9), and operation under deployment (11). Three major conclusions follow from the synthesis. First, algorithmic efficiency – including pruning, quantization, and sparsity might reduce computational burden to meet carbon goals; however, they only achieve concrete carbon reduction when accounted for in hardware and data centre setups. Second, life-cycle analyses show that while operational energy continues to account for most emissions during large-scale training, the embodied carbon from semiconductor fabrication plays an increasingly important role in fleets with a lot of equipment or frequent refreshes. Third, deployment decisions such as data center location, carbon-aware scheduling, and cloud–edge workloads placement bring much more variance on real emissions compared to what can be achieved at model-level optimisation. Between sectors, inconsistencies in methodology – notably for carbon reporting, system boundaries, and energy telemetry – hinder reproducibility and comparison of findings across studies. To overcome these limitations, this review suggests a research agenda with respect to standardized carbon accounting, hardware–software co-optimization, better available embodied-emission data, and the inclusion of carbon in decision-making for AutoML and scheduling systems. This comprehensive integration will serve as a stepping-stone towards the development of sustainable AI across academia and industry.</p>\u0000 </div>","PeriodicalId":538,"journal":{"name":"Energy Informatics","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2026-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1186/s42162-026-00651-8.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147561097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}