The successful reduction of carbon emissions in major sectors such as heavy industry and long-distance transport depends crucially on the ability to produce green hydrogen on a large scale. This involves generating hydrogen via water electrolysis, utilizing power sourced from renewable energies. However, persistent challenges, such as dynamic inefficiencies, material degradation, and renewable intermittency, demand a paradigm shift from static control strategies to adaptive, self-optimizing systems. This perspective argues that the synergistic integration of digital twins (DTs) and machine learning (ML) offers a transformative framework for real-time optimization, predictive maintenance, and resilient grid integration. By synthesizing physics-based modeling with data-driven intelligence, DT-ML systems enable closed-loop control architectures that dynamically adapt to operational uncertainties. We analyze the technical foundations of this integration, address critical barriers, and propose actionable pathways for stakeholders to accelerate the hydrogen economy's transition from promise to practice.

The key technologies of fuel cell systems have a great impact on their performance and lifetime. However, most of the current analyses in reviews only focus on a single system or one aspect of the technology. It is highly necessary to systematically analyze the new development and research trends of vehicular fuel cell systems. This paper proposes a quantitative and visual analysis based on bibliometrics to more intuitively reveal the research direction in this field. Key information on hydrogen/air subsystems, energy management strategies, and thermal management subsystems are analyzed by combining the literature analysis and technical discussion. In detail, the possible reasons behind the variations in the number of articles published within different periods of the subsystem are revealed. Then, through qualitative analysis of representative papers with high citations in different periods, the main research methods adopted by scholars in different periods are derived. Finally, the technology research trends of control methods that need to be addressed in the future are summarized for different subsystems. This paper provides a reference for researchers in the field to understand more clearly the current trends in the field and provide new ideas to break through the challenges of developing high-performance fuel cell systems.

In real-world driving scenarios, understanding the intentions of pedestrians in real-time is critical for the built environment safety when operating intelligent vehicles on the roads. Pedestrians crossing the street is a common behavior that can easily lead to accidents. This paper presents a comprehensive review of the prediction of pedestrian crossing intentions, focusing on data, model structure, data representation, information extraction, prediction function, and associated models and challenges. The review highlights that data types, model generalization ability, and prediction uncertainty are key challenges on pedestrian crossing intention prediction. It identifies open challenges and opportunities for future research in pedestrian crossing intention prediction.

To enhance the testing efficiency of autonomous vehicles, it is essential to derive intelligent traffic test scenarios. Current methods face limitations such as reliance on subjective analysis and neglect of inter-element correlations. This study introduces Kalman particle filtering theory for high-dimensional traffic scenario derivation. By analyzing comprehensive energy fields in normalized scenes with various elements, we define benchmark scenes using field energy theory. Multi-level research is conducted on processing high-dimensional spatial element data, proposing a normative paradigm for weight allocation among scene elements. We perform generalized derivation by extending hierarchical elements based on offset values, meeting functional verification requirements. Simulation experiments comparing risk event detection, decision-making, and feedback behavior between the proposed method and actual driving data show a steering matching index of 0.92, a longitudinal speed matching index of 0.96, and an root mean squared error (RMSE) mean value of 0.06.

Solid-state batteries have garnered global attention due to their high energy density and safety, holding the potential to replace traditional lithium-ion batteries in the electric vehicle, alleviating "range anxiety". However, a comprehensive solid-state battery management system to complement these batteries has not yet been systematically proposed. We attempt to construct a management system for solid-state batteries based on various characteristics, considering both the demand- and supply-side. This review first introduces the advantages and disadvantages of polymer-based, oxide-based, and sulfide-based solid electrolytes, as well as anode and cathode materials with higher specific capacities and promising applications. It lists the cycling performance and safety demonstrated by assembled solid-state pouch cells. Then, we systematically analyzes the differences between all-solid-state batteries and traditional lithium-ion batteries, proposing the demands and development directions for all-solid-state battery management from multiple dimensions. Finally, we build an all-solid-state battery management system from aspects such as signal monitoring, model building, aging, and early warning of failure, which includes three parts: the electric management system, the thermal management system, and the pressure management system. This review aims to provide researchers with new insights and ideas, promoting the efficient development and industrial upgrading of battery management technology.

Aqueous zinc-ion batteries have emerged as a promising energy storage technology due to their intrinsic safety, cost-effectiveness, and environmental sustainability. However, practical deployment of aqueous zinc-ion batteries faces critical challenges related to zinc anode instability, including dendrite formation, hydrogen evolution, corrosion, and interfacial degradation. This review highlights recent advancements on modulating interfacial chemistry and designing robust solid electrolyte interphases (SEI) layers enriched with functional components to address these challenges. Strategies including advanced electrode/electrolyte interface engineering, functional electrolyte additives, and artificial SEI are explored for improving zinc metal anode stability and overall battery performance. Additionally, cutting-edge in-situ characterization techniques across various scales and dimensions are introduced, offering unprecedented insights into the SEI formation processes, dendrite suppression, and Zn²⁺ transport dynamics. By integrating functional interfacial designs with advanced characterization, this work outlines innovative approaches to improve zinc anode performance and realize high-efficiency, long-lifespan aqueous zinc-ion batteries.

The accurate estimation of the remaining charge time (RCT) is essential in a battery management system (BMS), because it guarantees the safety and dependability of the power battery systems of new energy vehicles. However, the direct estimation of RCT is challenging because of the variability of actual charging scenarios and the complex charging process, which complicates the estimation of RCT in actual scenarios. Hence, this paper proposes an estimation framework based on deep learning for multi-scenario charging data to estimate the remaining charging times. Through an in-depth analysis of multi-scenario charging data, the RCT of the charging process is estimated using the temporal convolutional network (TCN) model, which has a strong generalization ability. Additionally, a dynamic learning rate (DLR) mechanism and an early stopping strategy (ES) are designed in the TCN model (DLR-ES TCN) for the nonlinear characteristics of the battery system to balance the relationship between model convergence speed and accuracy. Finally, compared with the traditional TCN model and four common deep learning models under three different scenarios, the experimental results show the mean absolute percentage error (MAPE) of the proposed method is less than 2%, indicating better accuracy and stability. This research can improve the safety monitoring of power batteries when applied to various target domains.

The emergence of electric vehicles (EVs) is critical for reducing environmental impacts and advancing sustainable transportation. Despite the known benefits, the complexities of EV integration with urban systems, power infrastructure, and environmental policies remain in-sufficiently explored. This literature review seeks to fill these gaps by synthesizing recent research across three key domains: environmental impact and sustainability, pattern analysis and system science, and methodology and operation of EV systems. The focus is on how EV adoption impacts climate change, net-zero emissions, air pollution, power grid dynamics, and mobility patterns, along with evaluating the effectiveness of AI-driven methodologies in enhancing EV operation. We conducted a systematic literature review of top-tier publications from journals such as Nature, Science, and PNAS, covering environmental science, urban systems engineering, and operational research related to EVs. Our findings highlighted that while EVs have the potential to reduce greenhouse gas emissions and improve urban air quality significantly, their actual environmental benefits depend heavily on factors such as electricity generation sources and the lifecycle impacts of battery production. Furthermore, we identified patterns in EV integration that can inform infrastructure planning and revealed the efficacy of advanced computational techniques, particularly reinforcement learning, in optimizing EV operations. Our major contributions include a comprehensive overview of the environmental impacts of EVs, insights into the challenges and opportunities of integrating EVs into urban systems, and showcasing the potential of AI-driven methodologies to enhance EV operations, with significant implications for policymakers, urban planners, and researchers.

The increasing reliance on batteries in transportation and energy storage sectors plays a pivotal role in addressing the challenges of energy security and grid power instability. However, the recurrent occurrence of battery safety incidents has emerged as the primary obstacle to their more extensive deployment. This necessitates the implementation of precise and efficient battery safety management technology to enhance the safety of batteries throughout their lifecycle, safeguarding the users’ assets and optimizing energy utilization. This article explores battery safety management technologies for power and energy batteries, starting with an overview of battery technology and then reviewing battery applications, failure mechanisms, and the analysis of existing intelligent safety management technologies. Finally, the paper consolidates current advancements, pinpoints gaps, and projects future trends in intelligent safety management technologies for power and energy-storage batteries. The insights presented will serve as a valuable reference and guideline for future research and development of battery safety management technology.

Digital twin technology used to realize the interactive mapping between digital model and physical entity in virtual space plays a crucial role in promoting the transformation of battery management to digitalization and intelligence. The key to achieving a digital twin is developing a virtual model that can accurately reflect the physical object. However, the intricate time-varying and nonlinear reaction characteristics within the battery often pose challenges in simulating complex operational conditions and maintaining high accuracy throughout the full life cycle. Incremental learning algorithms are suitable for online streaming data processing and can adapt to concept drift of the data stream by receiving new data online without retraining the entire model from scratch. This paper employs a simplified electrochemical model of the battery and integrates an Aggregated Mondrian Forest with incremental learning capabilities to construct a hybrid mechanism-data battery model. The developed hybrid model can acquire battery data and train in real time during battery operation to realize co-evolution with the physical battery to ensure the voltage prediction accuracy during the full life cycle of the battery.