The rapid growth of renewable energy sources, particularly solar and wind power, has increased the demand for efficient and reliable battery energy storage systems (BESSs) to address intermittency, enhance grid stability, and improve energy management. In recent years, artificial intelligence (AI) has emerged as a transformative technology for optimizing the operation, control, monitoring, and maintenance of battery storage systems. This review provides a comprehensive overview of AI-driven BESS technologies for renewable energy applications. The study examines recent advances in machine learning, deep learning, reinforcement learning, and hybrid intelligent algorithms applied to battery state estimation, energy management, fault diagnosis, predictive maintenance, thermal management, and lifetime prediction. Furthermore, the integration of AI-based BESSs with photovoltaic systems, wind farms, microgrids, and smart grids is critically analyzed. The review highlights the advantages of AI techniques in improving system efficiency, reliability, adaptability, and decision-making capabilities under uncertain operating conditions. Current challenges, including data quality, model interpretability, computational requirements, cybersecurity concerns, and real-time implementation issues, are also discussed. Finally, emerging research directions such as digital twins, explainable artificial intelligence, federated learning, and edge intelligence are explored to provide insights into the future development of intelligent battery storage systems. This review aims to serve as a valuable reference for researchers, engineers, and practitioners working at the intersection of artificial intelligence, battery technologies, and renewable energy systems.
Carbon-fiber structural batteries represent a class of multifunctional energy-storage systems that integrate electrochemical energy storage with mechanical load-bearing capability. Unlike conventional batteries, which are mainly evaluated based on cell-level energy density, structural batteries provide new opportunities for system-level weight reduction by reducing inactive structural mass, improving space utilization, and enabling distributed energy storage within integrated structures. In recent years, substantial progress has been achieved in carbon-fiber electrodes, structural electrolytes, laminated devices, electrolyte topology engineering, and fully carbon-fiber structural batteries. Nevertheless, most reported advances have been demonstrated under relatively ideal static testing conditions, while maintaining multifunctional performance under manufacturing and long-term service conditions remains a critical challenge. This review systematically examines the development of carbon-fiber structural batteries from a reliability perspective. First, the system-level motivations and technological evolution are introduced, and existing architectures are categorized according to their integration depth and degree of multifunctional coupling. Carbon-fiber electrodes are then discussed with emphasis on balancing capacity, ion transport, cycling stability, mechanical property retention, interfacial robustness, and manufacturing scalability. Furthermore, structural electrolytes are reviewed from the viewpoint of topology-enabled regulation of ion transport and load transfer, with particular focus on the intrinsic trade-off between ionic conductivity and mechanical modulus. In addition, manufacturing routes and device architectures are analyzed from the perspective of multifunctionality-degrading defects, including voids, dry regions, coating cracks, weak interfaces, and current-collector discontinuities. Finally, retained multifunctionality is used as a reliability-oriented evaluation criterion to examine the preservation of electrochemical, mechanical, interfacial, and safety functions, with particular emphasis on the carbon-fiber-specific failure chain linking interfacial and manufacturing heterogeneities to multifunctionality-degrading defects, coupled-field localization, and damage propagation. This review emphasizes that reliable carbon-fiber structural batteries require application-specific and coordinated optimization of materials, interfaces, electrolyte topology, coupled degradation behavior, and validation protocols.
Large volume expansions due to lithiation create problems for silicon anode mechanical stability, as volume changes apply large stresses to the electrode composite and copper current collector. The resulting mechanical damage includes current collector wrinkling or active material delamination, or both, which lead to capacity fade and current hotspots. Strategic design of the electrode microstructural properties, including controlling electrode porosity and adhesion, has the potential to help alleviate these issues. Two strategies to control microstructure are the choice of current collector and electrode drying conditions after casting; however, these strategies are not found to be sufficient to fully solve the damage caused by high stress. As a third approach, this paper investigates a different slurry formulation that uses less silicon content (75 wt%, compared to the usual 92 wt% active material) to limit anode mechanical damage. Reducing the silicon content of the anode eliminates delamination from high-strength current collectors, improving electrochemical cycle lifetimes significantly. However, lower silicon content leads to potential losses in energy density, which we address using a multilayer architecture comprising the 75 wt% Si layer as an interface between the current collector and the 92 wt% active material layer; this interfacial layer prevents delamination while maximizing energy density. Despite improvement, minor wrinkles remain evident even with the multilayer architecture, indicating the need to combine this work with additional mitigations in the future.
This study aims to verify and improve the Electrical Equivalent Circuit Model (EECM) for a regenerated VW e-Golf cell and to develop a verification and optimization framework that enhances simulation accuracy. The model is based on an identified set of EESB elements derived from enhanced measurements of the regenerated cell and is compared with the original cell. A global EESB model is implemented in the PLECS environment, comprising a charge/discharge block and a Voc versus SOC evaluation. Parameters are obtained from measurements of the regenerated VW e-Golf cell and augmented with SOC-dependent polynomial relationships for individual model components. The methodology was applied to identify EESB elements for the regenerated VW e-Golf cell and to produce an EESB model aligned with the identification results. Verification compares simulated and experimental curves in critical SOC regions (0–10%, around 30%, and during relaxation) and cross-validates regenerated versus original cells. Results show that SOC-based polynomial estimates extend the valid range of EESB elements to 0–10% SOC and improve agreement with measured trajectories. Optimization reduces the computational load and improves accuracy, particularly in critical SOC regions, supporting a robust verification framework for regenerated battery cells and guiding further research and implementation in BMS and simulation environments.
Thermal runaway (TR) severity can be ranked using absolute or energy-normalised heat and by the pathway through which heat leaves the cell, but whether these definitions identify the same high-hazard cells remains unclear. We analysed 98 non-ISC heater-triggered tests from 13 commercial or near-commercial cylindrical cell models in the Battery Failure Databank. Cathode chemistry was not analysed because it was incompletely documented. Total, body, and ejecta heat were evaluated using allometric scaling, S-FTRC-generation-adjusted fixed-effects models, hierarchical bootstrap ranking, and isometric log-ratio analysis. Scaling exponents for total, ejecta, and body heat were 0.899, 1.003, and 0.879, respectively, with all 95% bootstrap intervals including b = 1. Despite this approximate proportional scaling, generation-adjusted absolute-total and total-kJ/Wh rankings disagreed for 46.2% of cell-model pairs. Body- and ejecta-heat rankings diverged even more strongly, with 62.8% point discordance, a bootstrap median of 61.5% (95% interval 53.8–67.9%), and 28 of 78 pairs showing reversal probabilities ≥95%. These findings show that absolute heat, energy-normalised heat, body heat, and ejecta heat address different engineering questions and should be reported and interpreted separately rather than combined into a universal cell-safety ranking.
We propose a two-stage joint estimation method combining a multi-physics-constrained physics-informed neural network (PINN) with an adaptive extended Kalman filter (AEKF) for lithium-ion battery state-of-charge (SOC) estimation. Three constraints—an RC polarization dynamics ODE residual, discharge voltage–SOC monotonicity, and terminal-voltage physical bounds—are embedded into the PINN loss function; the converged network then serves as the nonlinear observation model within the AEKF. On LG 18650HG2 cells across six temperatures (−20 to 40 °C) under a strict cross-condition setup (training on LA92/UDDS; testing on US06 and two mixed profiles), the method maintains a temperature-averaged SOC mean absolute error (MAE) of 2.33% (1.54–4.48%, three random seeds), whereas the MAE for the equivalent circuit model with the AEKF (ECM-AEKF) degrades to 9.46% and 11.66% at −20 °C and −10 °C, and remains 1.5–4.6× worse with the per-temperature re-identified parameters; an end-to-end long short-term memory (LSTM) baseline averages 1.90% but degrades to a per-condition maxima of 22.6% at low temperatures. Multi-seed ablations locate the constraints’ value in providing low-temperature stability and physical consistency, with the RC-ODE constraint contributing most. The study further reveals that goodness of terminal-voltage fit does not imply SOC accuracy; the controlled comparisons trace this to the model structure rather than parameter settings, exposing the risk of voltage-fitting-based evaluation over wide temperature ranges.
Commercial polyolefin separators for lithium-ion batteries (LIBs) exhibit only inadequate wettability and thermal stability. In large-scale production, high electrolyte uptake and wetting are essential to enable rapid electrolyte filling during battery assembly to reduce costs. In addition, it is imperative to develop separators with enhanced thermal stability for improved performance and safety. The focus of this study is the structure–property–performance relationship of separator coatings. Platelet-shaped glass particles are utilized as inorganic coating material for polyethylene (PE) separators. Styrene–butadiene rubber (SBR) was selected as binder due to its high thermal stability. Hybrid separators are prepared using a colloidal coating technology employing a water-based slurry. As the excessive use of binder in the coating can block pores, precise control of the binder content was essential to maintain battery performance. The resulting separators with an optimized binder content of 1 wt.% in the coating demonstrate high porosity, instantaneous wetting with electrolyte, and a 25 K increase in onset temperature for shrinkage. The utilization of glass platelets with an aspect ratio of 10 as coating material provided the best balance among processability, coating homogeneity, thermal stability, ionic conductivity, and cycling performance under the investigated processing conditions.
Sulfide solid electrolytes (SSEs) are promising for all-solid-state lithium batteries (ASSLBs) due to their high ionic conductivity, mechanical deformability, and interfacial compatibility. However, SSE interfaces with anodes, cathodes, conductive additives, and current collectors are unstable, triggering safety failures like capacity degradation, internal resistance build-up, thermal runaway, and short circuits. This review summarizes recent progress on interface-induced safety failure mechanisms in sulfide-based ASSLBs, focusing on interface types, failure mechanisms, and thermal/mechanical degradation under multi-field coupling. We survey interface modification strategies and highlight advanced characterization techniques for probing interfacial phenomena. Key challenges and future research directions are discussed. Integrating recent findings, we identify interfacial instability as the primary bottleneck governing safety failures, providing a theoretical and technical framework for rational interface design, performance optimization, and safety enhancement. Throughout this review, we use SSE as the standard abbreviation for sulfide solid electrolytes.
Long-term degradation of vanadium redox flow batteries (VRFBs) is strongly coupled with the evolution of discharge direct current internal resistance (DCIR). This work employs 213 constant-current cycles on a 4 cm2 single cell to examine the associations between discharge DCIR and capacity, discharge voltage and efficiencies. The average values of the initial 10 cycles were used as the baseline. Multiple statistical approaches, including detrending, differencing, moving-block bootstrap, and internal chronological holdout evaluation were used to assess the influence of shared temporal trends and serial dependence. Self-calculated DCIR matches test records closely with merely 1.03% average relative error. After trend correction, normalized DCIR maintains a strong negative correlation with normalized capacity. The established free-intercept quadratic model achieved the best full-data fitting performance, with an R2 of 0.99593 and a root mean square error (RMSE) of 0.00524. The fitted empirical relationship indicates that equal increments in normalized DCIR are associated with larger concurrent capacity-state reductions in the higher-resistance region. During the internal chronological holdout evaluation, the DCIR-based model yielded an RMSE of 0.00854. These results establish a statistically robust DCIR–capacity relationship over long-term cycling and demonstrate the potential of routinely recorded discharge DCIR as a low-cost concurrent capacity-state indicator, providing a quantitative foundation for its future extension to broader VRFB operating scenarios.
Accurate and robust state-of-health (SOH) estimation is essential for efficient battery management systems (BMS), especially in practical scenarios suffering from severe cell-to-cell inconsistencies and data distribution shifts. Although prevailing data-driven estimation methods can achieve satisfactory accuracy within specific domains, their generalization capability degrades drastically when applied to unseen battery cells. To fill this research gap, this paper develops an integrated physics-guided and domain-aware SOH estimation framework. The proposed framework combines redundancy-aware feature screening, cycle-aware soft covariance alignment specifically designed for physics-based health indicators, as well as monotonicity-constrained LightGBM regression embedded with split conformal uncertainty quantification. Experimental validations are conducted on the NASA B0005, B0006 and B0007 battery datasets under the leave-one-cell-out (LOCO) cross-cell evaluation strategy. Comparative results reveal that the presented method achieves prominent performance improvement on the most difficult B6 domain, where the root mean square error (RMSE) is reduced from 0.1068 to 0.0845 with a decline rate of 20.9%, and the coefficient of determination (R2) rises from 0.2464 to 0.5286, while maintaining stable estimation accuracy on less challenging target domains. In terms of overall performance, the average RMSE across all tested cells drops from 0.0499 to 0.0425, corresponding to a 14.8% reduction, indicating improved cross-cell performance across the investigated cells. Furthermore, the adaptive alignment mechanism can be dynamically activated only when necessary, effectively avoiding redundant distortion of the original feature distribution. The research findings indicate that the integrated framework can alleviate cross-cell battery degradation discrepancies under the investigated conditions. The proposed strategy demonstrates promising potential for improving cross-cell SOH estimation under the investigated laboratory conditions; however, the uncertainty intervals are not fully calibrated under severe domain shift, and broader validation on larger and more heterogeneous battery datasets is required before practical deployment claims.
Battery-management systems observe current, terminal voltage, limited temperature measurements, and operating setpoints, but not the internal variables of electrochemical theory, so routine signals generally cannot identify a unique mechanism. We present a human-guided, generative-AI-assisted methodology that translates physics-based expectations into deterministic, auditable screening rules: human scientific authority fixes the physical assumptions, evidence requirements, and permissible claims, artificial intelligence supports development, and runtime evaluation is non-generative. LiFePO4 is the test case. A Zeng–Bazant current-dependent plateau approximation supplies a physics-based reference, and a bivariate representational precedent motivates a composite, reference-dependent voltage residual that is not identified as thermodynamic work. Eleven observable screens return present, absent within resolution, indeterminate, or unavailable. Four evidence forms are separated. Digitized model curves show the reduced plateau relation tracks its parent phase-field simulation through moderate rates, with a high-rate limitation. Published temperature-conditioned discharge profiles show stable plateau elevation and flattening from 268 to 298 K across 0.5 C–2 C, with mixed 2 C curvature. Measured replicates of a commercial cylindrical cell at two ambient setpoints resolve a within-run surface-temperature depression whose integrated first-law balance is heat-rejection-dominant and compatible with, but not uniquely attributed to, a literature-bounded reversible contribution. Controlled synthetic cases verify deterministic feature recovery and abstention without establishing a mechanism.
Solid-state sodium metal batteries are promising for large-scale energy storage owing to their high safety, abundant sodium resources, and low material cost. However, NASICON-type solid-state electrolytes, such as Na3Zr2Si2PO12, still suffer from limited room-temperature ionic conductivity and poor ceramic densification. Herein, a multivalent co-doping strategy using Zn2+, Sc3+, Hf4+, and Nb5+ was employed to regulate the crystal structure, sintering behavior, and Na+ transport properties of NASICON electrolytes. The effects of isovalent and aliovalent dopants were systematically investigated by XRD, Rietveld refinement, SEM-EDS, bond-valence-site calculations, and electrochemical impedance spectroscopy. The optimized monoclinic Na3.167Zn0.167Hf0.167Zr1.5Nb0.167Si2PO12 electrolyte delivered a room-temperature total ionic conductivity of 1.16 mS cm−1 and an activation energy of 0.35 eV, mainly due to optimized Na+ migration-channel geometry and enhanced ceramic densification. Na||Na symmetric cells exhibited stable cycling for 500 h with a maximum polarization voltage of 20 mV. Furthermore, solid-state Na||Na3V2(PO4)3 cells retained 95% capacity after 624 cycles at 1 C. This work provides a feasible doping strategy for advanced NASICON electrolytes and solid-state sodium metal batteries.
This study involved connecting four 18650 cylindrical lithium-ion batteries in series. An ABS structure produced using a 3D printer was then filled with salt hydrate phase change material (PCM). The batteries were then tested in two separate systems: one containing PCM and one without. In the PCM-enhanced system, the maximum surface temperature did not exceed 46 °C during discharge at a constant current of 8 A, corresponding to a discharge rate of 2.85 C. In the PCM-free system, the maximum surface temperature reached 66.3 °C under the same discharge condition. This resulted in improved battery performance by reducing temperature rise and enhancing discharge behavior under high-rate operating conditions. Furthermore, using PCM increased the batteries’ energy output by 20% during discharge at a constant 2.85 C rate. Finally, a techno-economic and environmental analysis was conducted to assess the system’s suitability for micro-mobility applications with a 1 kWh capacity. The system reached the break-even point in 4 years and 2 months, indicating that it is economical and feasible.
This work develops a hybrid kernel sparse Gaussian process regression integrated with kernel density estimation (HCSGPR-UQ) to resolve three critical drawbacks of conventional lithium-ion battery State of Energy (SOE) estimators: degraded accuracy under dynamic loads, high computational overhead, and inadequate uncertainty quantification. A composite covariance kernel is built by weighting the radial basis function (RBF) and Matérn 5/2 kernels to simultaneously model global smooth SOE decay trends and local nonlinear fluctuations induced by abrupt current/temperature variations. Inducing-point sparse approximation is adopted to accelerate model inference, while kernel density estimation (KDE) generates nonparametric prediction bounds for quantitative uncertainty evaluation. Validations are carried out on a 75 Ah traction lithium-ion cell across −5 °C to 35 °C under Dynamic Stress Test (DST) and Beijing Bus Dynamic Stress Test (BBDST) cycles. Experimental results reveal that the proposed method yields mean absolute errors (MAEs) of only 0.32% (DST) and 0.38% (BBDST), runs roughly 15× faster than full Gaussian process regression (GPR), and attains a 94.7% coverage probability for nominal 95% prediction intervals. Balancing estimation precision, real-time inference speed and statistical reliability, the proposed framework delivers a viable online SOE estimation solution for vehicle battery management systems (BMSs).
With the widespread application of lithium-ion batteries in electric vehicles, degradation diagnosis has attracted increasing attention. In practical operating scenarios, however, path-dependent degradation induced by the alternating effects of calendar aging and cycling aging can significantly influence the diagnosis of battery degradation. Existing methods often identify degradation under these two aging conditions in isolation, making it difficult to quantify their coupled impact. To address this issue, this study applies a physically constrained half-cell OCP reconstruction framework to quantify electrode-level degradation under coupled aging conditions. Specifically, degradation parameters are identified by fitting full-cell pseudo-open-circuit voltage (pOCV) curves with half-cell open-circuit potential (OCP) profiles, and the corresponding degradation modes are further quantified. The results show that the proposed model can reconstruct full-cell pOCV under different aging conditions with an RMSE maintained below 10.5 mV. The loss of active material in the anode is highly sensitive to continuous cycling, whereas the divergence of loss of lithium inventory under alternating aging conditions is relatively weak but shows pronounced differences under distinct single-aging conditions. This method enables electrode-level diagnosis of path-dependent degradation under alternating aging conditions and provides a foundation for reliable degradation diagnosis under complex operating scenarios.
The electrification of commercial vehicles demands precise battery thermal management, but direct measurement of the cell core temperature is challenging. This paper presents an electrochemical impedance spectroscopy (EIS)-based approach for rapid indirect estimation of the mean internal temperature in 2170 NMC lithium-ion cells. Three measurement approaches and various fitting methods, including Steinhart–Hart, least-squares polynomials, and nonlinear Arrhenius-based fits, are compared using experimental data. The results indicate that estimation accuracy is more strongly influenced by the selection of measurement frequency than by the choice of fitting approach. The optimal method combines a single optimized frequency with a nonlinear polynomial incorporating an Arrhenius term, achieving a maximum deviation of 0.23K and a mean deviation of 0.14K. This framework enables indirect EIS-based estimation of the cell core temperature and can be further refined through measurements on multiple cells or by combining different fitting methods. Future work should extend the proposed approach to dynamic EIS measurements, battery ageing, and integration with thermal models for early overheating warning and identification of the first thermal runaway warning stage in high-power commercial vehicles.
Fault detection from electric vehicle charging segments is challenging because real-world records are noisy, verified fault labels are scarce, and weak signatures may evolve gradually across time and unevenly across a vehicle’s charging history. This study proposes an unsupervised dynamic multiscale state-space support vector data description framework for vehicle-level battery fault detection. A gated diagonal state-space encoder preserves long-range charging patterns while retaining a direct input path for transient changes. A progressive cross-scale fusion module then combines short-term fluctuations with accumulated deviations in the learned hidden representation. Finally, a two-stage hypersphere optimisation strategy first estimates the normal centre and then refines the boundary around that fixed centre. This coordinated design avoids sequence reconstruction and directly scores charging segments by their distance from normal behaviour before robust vehicle-level aggregation. On the two EVBattery subsets that permit statistically reliable evaluation, the proposed framework achieved vehicle-level areas under the receiver operating characteristic curves of 0.8849 and 0.8438. These results exceed those of the best-performing baseline on the corresponding subsets by 0.0889 and 0.0417, respectively. The results show that coordinating long-range encoding, cross-scale fusion, and staged boundary learning improves threshold-independent vehicle-level fault ranking in real charging data.
Solid electrolyte interphase (SEI) characterization is needed to interpret the performance, degradation, and lifetime of graphite and Si-containing carbon-based negative electrodes in lithium-ion batteries. However, SEI claims are often difficult to compare because measured signals, inferred assignments, sample history, and electrode architecture are not always clearly separated. This review presents a claim-bounded framework for SEI characterization that distinguishes direct observables from inferred chemical, molecular, structural, morphological, and functional information. Photoelectron spectroscopy methods provide chemical-state and relative-depth-sensitivity constraints; secondary-ion mass spectrometry methods provide fragment and isotope distributions; vibrational spectroscopies support functional-group and local vibrational evidence; nuclear magnetic resonance and molecular mass spectrometry provide molecular or product-level constraints; and microscopy, tomography, and atomic force microscopy provide morphology, architecture, local thickness, topography, and mechanical response. Across these methods, rinsing, drying, sputtering, beam exposure, extraction, and limited sampling can alter the observable and therefore the defensible claim. The review emphasizes the distinction between native electrode-associated SEI features and extracted, soluble, or electrolyte-phase products, and between morphology-only evidence and chemically assigned morphology. It concludes by proposing claim-driven correlative workflows and reporting guidance for reproducible interpretation on graphite, Si/graphite, Si/C, and carbon-coated Si architectures where directly studied or present.
Lithium-ion battery fires are governed by continuous heat release during thermal runaway, flammable gas venting, and thermal coupling between adjacent cells. Even after visible flames are extinguished, post-extinguishment temperature rise, thermal runaway propagation, and reignition may still occur. This review establishes a full-process control framework linking flame suppression, sustained cooling, thermal runaway propagation mitigation, and reignition control. Within this framework, water-based agents, clean gaseous agents, dry powder agents, foams, cryogenic media, and hybrid suppression methods are not only compared by their flame extinguishment performance but also by their cooling capability, thermal runaway propagation/reignition control, scenario applicability, and environmental impacts. Evaluation metrics and standardization requirements are further integrated to support cross-study comparison and practical suppressant selection. Existing studies indicate that a single suppressant is generally unable to achieve both rapid flame extinguishment and post-extinguishment thermal stability. Multi-mechanism synergy, realistic scenario validation, and standardized evaluation protocols are therefore essential for improving the full-process control of lithium-ion battery fires.
Lithium metal batteries are widely regarded as one of the most promising candidates for achieving energy densities beyond 500 Wh kg−1. However, their practical commercialization is severely hindered by the high reactivity of lithium metal, which leads to pronounced interfacial instability. Constructing an artificial solid electrolyte interphase (SEI) via surface pretreatment has been demonstrated to be an effective strategy for suppressing dendrite growth and mitigating parasitic side reactions. Herein, we take advantage of the rapid reaction between lipoic acid (LA) and lithium metal to pre-form a uniform artificial SEI on the anode surface. This interphase is composed of organic COO-Li species and sulfur-containing compounds. Electrochemically, the LA-modified lithium anode exhibits remarkable stability, sustaining more than 1500 h of cycling in symmetric cells at 5 mA cm−2 and 5 mAh cm−2. Furthermore, full-cell configurations, including lithium-sulfur and lithium-LiFePO4 pouch cells, deliver significantly improved cycling performance compared with those employing bare lithium anodes. These results establish a practical and scalable route for fabricating stable artificial SEI layers on lithium metal, thereby providing a feasible pathway toward the realization of high-energy-density lithium metal batteries.
