中文核心期刊
中国科技核心期刊
中国化学与物理电源行业协会会刊
中国电子学会化学与物理电源分会会刊
Home Table of Contents

20 September 2025, Volume 49 Issue 9
    

  • Select all
    |
    Smart Battery System
  • Application status and prospects of artificial intelligence in battery technology
    BAI Wanlong, WU Jingyu, XIANG Yu, DING Fei, YAN Yiming
    Chinese Journal of Power Sources. 2025, 49(9): 1784-1790. https://doi.org/10.3969/j.issn.1002-087X.2025.09.001
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Battery technology, as the central pillar for efficient storage and utilization of clean energy, confronts several critical challenges—prolonged research‑and‑development cycles, difficulty in accurate lifetime forecasting, insufficient safety assurances, and low recycling efficiency. Artificial intelligence, through the construction of data‑driven models and intelligent optimization algorithms, has accelerated advancement across multiple stages. Looking ahead, a closed “physical entity—virtual mapping—intelligent decision” loop centered on digital twins, augmented by explainable AI and lightweight algorithms, promises end‑to‑end coverage from molecular design and manufacturing optimization to online monitoring and second‑life reuse. Realizing this intelligent‑fusion paradigm in practice will require the establishment of shared data platforms to break down data silos and demystify model black boxes, as well as strengthened interdisciplinary collaboration—thereby delivering high safety, low cost, and sustainable battery systems to propel the intelligent transformation and green development of the new energy sector.
  • A review of methods for constructing fractional-order models of on-board power batteries
    WU Longxing, WEI Xinyuan, SUN Peng, LIU Chunhui
    Chinese Journal of Power Sources. 2025, 49(9): 1791-1800. https://doi.org/10.3969/j.issn.1002-087X.2025.09.002
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    With advancing automotive electrification, battery management systems (BMS) are crucial for intelligent electric vehicle (EV) operation. However, accurate state estimation relies on effective battery model selection. Fractional-order models (FOM) show promise, balancing computational efficiency and accuracy. Therefore, this paper primarily reviews the modeling of FOMs. First, this paper outlined the modeling mechanisms of different battery models. It then focused on the principles, structures, advantages, and application scenarios of FOMs compared to others. Next, the potential of FOM-thermal model coupling for precise, synergistic monitoring of battery temperature and electrochemical performance was explored. Finally, future FOMs research directions and challenges in battery management were discussed. Overall, this review aims to provide a reference for FOMs theoretical research and engineering applications. It seeks to promote their integration into advanced BMS.
  • Advances in element doping for stabilizing sodium-based layered oxide cathodes
    MAO Mian, ZHANG Juncai, QIAO Yaru, ZHANG Zhuangzhuang
    Chinese Journal of Power Sources. 2025, 49(9): 1801-1812. https://doi.org/10.3969/j.issn.1002-087X.2025.09.003
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Sodium-ion batteries (SIBs) have broad application prospects in the field of large-scale energy storage. Layered oxide cathodes (LOCs), due to low cost, high energy density, and mature preparation process, have become ideal cathode materials for SIBs. However, in practical applications, LOCs faced many issues such as irreversible phase transitions, migration and dissolution of metal cations, and severe interfacial side reactions, which led to the degradation of battery performance. In response to these challenges, researchers carried out extensive work and made remarkable progress especially in microscopic local electronic regulation. A detailed summary of the scientific issues, that urgently need to be addressed regarding LOCs, was provided, the research advancements in these fields were outlined in recent years, the mechanism of how the regulation of local electronic structure affects the stability of LOCs was further explored, and aiming to offer valuable insights for improving the stability of SIBs cathode materials, thereby promoting the development and application of SIBs.
  • Research progress on influence of environmental stress on lithium ion batteries in electrochemical energy storage systems
    HU Zhenkai, PENG Peng, SUN Wanzhou, TAN Qipeng
    Chinese Journal of Power Sources. 2025, 49(9): 1813-1823. https://doi.org/10.3969/j.issn.1002-087X.2025.09.004
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Lithium ion battery electrochemical storage systems are essential solutions for addressing the unpredictability of renewable energy and achieving high levels of energy consumption. The research investigates the impact of key environmental factors in actual service conditions on the performance and safety of lithium ion batteries. It systematically analyzes the effects of environmental stresses such as temperature, humidity, salt spray corrosion, pressure, and vibration on battery materials, capacity degradation, service life, and operational safety. The results indicate that extreme environmental stresses accelerate the degradation and aging of energy storage lithium ion batteries, posing significant challenges to their service life and long-term safety. In response to the threats posed by extreme environments, the research summarizes unresolved issues regarding lithium ion batteries in specialized environments and highlights effective strategies from existing research aimed at mitigating environmental impacts on battery reliability and safety. Furthermore, it identifies key directions for future research, particularly in improving the environmental adaptability of batteries and extending their lifespan. The research provides insights into the failure mechanisms of lithium ion batteries and offers a framework for designing energy storage systems suitable for applications in extreme environments.
  • Research progress on EIS state estimation and safety warning for lithium ion batteries
    WANG Dongdong, YANG Fahu, SHEN Haichao, QUAN Chaoming, HOU Zhengjian, MAO Deyuan, ZHANG Shengjuan, WANG Jianhua
    Chinese Journal of Power Sources. 2025, 49(9): 1824-1830. https://doi.org/10.3969/j.issn.1002-087X.2025.09.005
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Electrochemical impedance spectroscopy (EIS), as an electrochemical testing technique, measures the impedance response of a battery by applying a small-amplitude AC excitation signal, which provides detailed information about the internal structure and chemical reactions of the battery. In this paper, three applications of EIS-based testing for lithium ion battery state-of-charge (SOC) estimation, lithium ion battery state-of-health (SOH) estimation, and lithium ion battery safety warning are reviewed. The analysis and summary of the existing research work aims to provide reference and reference for the development of state estimation and safety early warning technologies for lithium ion batteries.
  • High performance zinc air battery based on enhanced adaptive hydrogel electrolyte
    CHEN Nuo, LAI Jingning, HUANG Yongxin, CHEN Renjie
    Chinese Journal of Power Sources. 2025, 49(9): 1831-1839. https://doi.org/10.3969/j.issn.1002-087X.2025.09.006
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Solid-state zinc-air batteries (ZABs) exhibit great potential for applications in flexible energy storage systems. In this study, an enhanced adaptive neutral hydrogel electrolyte was fabricated by incorporating sodium hyaluronate (SA), polyvinyl alcohol (PVA) and zinc salts using freeze-thaw cycling method. The physicochemical properties and electrochemical performance of the resulting material were systematically analyzed. The results demonstrate that the strong hydrogen bonding between SA and water molecules endows the material with excellent adaptive mechanical properties, water retention capability, and thermal stability. Additionally, the formation of hydrogen bonds effectively reduces water activity, thereby inhibiting hydrogen evolution side reactions and zinc dendrite growth. Moreover, the strong binding energy between SA and Zn2+ significantly alters the Zn2+ transport pathway. Benefiting from this optimized design, the symmetric battery exhibits stable cycling for over 5 000 hours at current density of 0.4 mA/cm2. Furthermore, the cycling lifespan and full discharge capacity of the ZABs are significantly improved.
  • Insight into the influence of structure degradation on battery safety stability for ternary cathode
    ZHANG Yan, ZHANG Ruiqiang, CHU Yinxiao, MENG Fanbo, XU Ting, SUN Xiao, GUO Jian
    Chinese Journal of Power Sources. 2025, 49(9): 1840-1845. https://doi.org/10.3969/j.issn.1002-087X.2025.09.007
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The poor thermal stability and detrimental gas release characteristics of ternary cathode materials significantly affect the safety performance of lithium ion batteries. Aimed at exploring the intrinsic relationship between performance failure, structural failure, and safety stability of the ternary cathode, the in-situ electrochemical mass spectrometry and differential scanning calorimetry analyses were conducted on the aged (A-NCM) and fresh (O-NCM) LiNi0.5Co0.2Mn0.3O2 cathodes to investigate their safety stability. The results show that, under an upper cutoff voltage of 4.7 V, significant CO2 and H2 harmful gas signals were detected for A-NCM during electrochemical cycling. Additionally, when the thermal decomposition reaction between the electrode and the electrolyte was initiated, A-NCM electrode would release more heat, further increasing the battery temperature and posing safety hazards. This research helps to construct the relationship between "battery aging and battery safety", providing a theoretical basis for intelligent battery safety management.
  • Preparation of nitrogen doped mesoporous carbon materials and its application in lithium sulfur batteries
    YU Kai, LIU Yunan, MI Juan, SUN Qiang
    Chinese Journal of Power Sources. 2025, 49(9): 1846-1851. https://doi.org/10.3969/j.issn.1002-087X.2025.09.008
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The study developed a method for preparing nitrogen doped mesoporous carbon impregnated sulfur composite cathode materials based on polydopamine coated nickel template method, targeting key issues such as poor electronic conductivity of active sulfur and discharge product lithium sulfide in lithium sulfur batteries, and rapid capacity decay caused by inherent shuttle effect of lithium polysulfide. By regulating the content of dopamine precursors, C-300 and C-600 nitrogen doped mesoporous carbon materials with differentiated pore structures were prepared. Structural characterization shows that C-300 material has higher structural defect density (ID/IG=2.82), larger specific surface area (798 m2/g), and pore volume (1.83 cm3/g), which are increased by 57.7%, 57.7%, and 96.8% respectively compared to C-600 material. Electrochemical analysis shows that the optimized multi-level pore structure enables the S@C-300 nitrogen doped mesoporous carbon/sulfur composite cathode material to exhibit excellent cycling stability at a rate of 0.5 C. After 100 cycles, the discharge specific capacity still maintains 850 mAh/g, with a capacity decay of 0.403% per-cycle, while maintaining nearly 100% coulombic efficiency.
  • Application of in-situ characterization technique for “self-enhanced” lithium-oxygen battery systems
    YANG Dong, ZHANG Shifeng, GAO Rui, LIU Xueling, DU Ruijie, WANG Liang, LI Yongli
    Chinese Journal of Power Sources. 2025, 49(9): 1852-1862. https://doi.org/10.3969/j.issn.1002-087X.2025.09.009
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    As the demands for battery capacity and safety increased, developing novel battery systems beyond conventional Li-ion batteries has emerged as a key research topic. “Self-enhanced” lithium-oxygen battery is a novel system that can automatically realize activity adjustment through responding to changes of the applied potential during the charge/discharge process. The exploration of in-situ characterization techniques for “self-enhanced” lithium-oxygen battery is essential for revealing reaction mechanisms and designing a better future smart battery system. In this paper, the application of in-situ characterization methods for “self-enhanced” Li-oxygen battery has been reviewed in detail. At the same time, this review also provides a future perspective of in-situ characterization methods for reaction mechanism explanation.
  • Research of bipyridinium-based single redox flow batteries technology
    CHEN Junwei, HUANG Guozhi, PENG Siran, WEN Xiankui, FAN Qiang, HU Quan, LI Chaojie, LI Wenjin, WANG Chao
    Chinese Journal of Power Sources. 2025, 49(9): 1863-1867. https://doi.org/10.3969/j.issn.1002-087X.2025.09.010
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Electroactive organic materials (EOMs) have garnered widespread attention due to their superior solubility, low cost, and synthetically tunable. Herein, based on the solubility statistics from the EPI Suite™ database, outstanding EOMs candidates are screened and designed for bipyridinium based single redox flow battery (SRFB), enabling the selected 2,2'-bipyridinium electrolyte to achieve stable charge/discharg process at saturated concentrations. Thanks to this design, a high reversible capacity of 2,2'-bipyridinium-based electrolytes has been realized. The SRFB based on saturated concentration of 2,2'-bipyridinium exhibits a reversible electrolyte capacity of up to 56.0 Ah/L at a current density of 40 mA/cm2, the coulombic efficiency is 99.99%. This study significantly enhances the reversible capacity of bipyridinium-based electrolytes, providing a promising and safe RFB technology that can be adapted to various application scenarios.
  • Application research on gradient thermal design in development of thermal battery
    QIANG Shanshan, ZHENG Xia, JI Liuyan, LUO Chongxiao, LU Zihang, BAI Yunfei
    Chinese Journal of Power Sources. 2025, 49(9): 1868-1872. https://doi.org/10.3969/j.issn.1002-087X.2025.09.011
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Aiming at the problem of “hot in the middle and cold at both ends” in the thermal battery stack with traditional single thermal ratio at the end of operation, a design scheme was proposed for a gradient thermal design of the stack. The design ensured that the heat production at both ends of the thermal battery stack was large, while the heat production in the middle of the stack was small, forming a temperature gradient. In addition to transferring heat to the metal casing, heat also slowly transferred to the middle. During low-temperature operation, thermal compensation was used to extend the thermal life of the battery; during high-temperature operation, controlling the heat production in the middle could control the side reactions of the thermal battery and improve the working time and safety. This could provide a reference for the design and development of similar thermal batteries.
  • Neural network ensemble method for branch current estimation of parallel battery system
    PEI Lei, YANG Jiawei, WANG Tiansi, LI Huanhuan
    Chinese Journal of Power Sources. 2025, 49(9): 1873-1880. https://doi.org/10.3969/j.issn.1002-087X.2025.09.012
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In a parallel battery module (PBM), cost and volume constraints preclude the installation of sensors to directly measure the current in each individual branch. However, due to inevitable inconsistencies among individual battery cells, and continuous deterioration caused by unbalanced temperature and aging further exacerbates this inconsistency, the maximum current that individual cells may withstand can reach several times the design value, posing significant hidden dangers to battery state analysis and safety management. To address this challenge, this paper systematically analyzes the influence of branch parameters on the overall behavior of PBMs. In addition to existing parameters such as voltage, current, and charge changes, a new characteristic parameter, “accumulated voltage”, has been introduced. Based on these selected parameters, a novel neural network ensemble method is designed. The effectiveness and universality of the new scheme are verified through experiments conducted under different numbers of cells and different discharge profiles. The estimation error is consistently controlled less than 1%.
  • Fault diagnosis method of lithium ion battery based on TTAO-VMD and LOA analysis
    LIAO Li, LI Xingke, WANG Yi, HUANG Yang, ZHENG Quanxin, JIANG Jiuchun
    Chinese Journal of Power Sources. 2025, 49(9): 1881-1889. https://doi.org/10.3969/j.issn.1002-087X.2025.09.013
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Lithium ion batteries are widely used in electric vehicles, and their thermal runaway faults and serious inconsistent faults have become serious safety hazards. In this paper, a battery fault diagnosis method based on Trigonometric Topology Aggregation Optimization (TTAO) algorithm to optimize Variational Mode Decomposition (VMD) combined with Longitudinal Outlier Average (LOA) analysis is proposed. Firstly, TTAO is used to optimize the modal number K and penalty factor a in VMD to improve the accuracy and stability of signal decomposition. Then, the first two Intrinsic Mode Functions (IMF) were extracted as fault features, and the voltage anomaly detection mechanism was constructed by combining the sliding window and LOA method, and the early warning of thermal runaway faults and the accurate identification and location of seriously inconsistent faults were realized with the help of threshold judgment strategy. Experimental results verify the robustness and reliability of the proposed method based on real vehicle operation data, and show higher robustness and lower false positives than traditional correlation coefficient methods.
  • Research on the design and optimization of a biomimetic leaf-vein-shaped liquid cooling plate for lithium ion batteries
    CHEN Kang, CAO Yuchun, REN Zhaoyong
    Chinese Journal of Power Sources. 2025, 49(9): 1890-1898. https://doi.org/10.3969/j.issn.1002-087X.2025.09.014
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To address the challenge of thermal runaway in lithium ion batteries during high-rate discharge, this study presents a battery cooling system incorporating a liquid cooling plate with veined flow channel (LCP-VC). Using numerical simulation, the cooling performance and fluid dynamic characteristics of the system at 5 C discharge were systematically analyzed. Four liquid cooling plate configurations-vein-shaped, serpentine, straight, and honeycomb-were compared. The results show that the maximum temperature (Tmax) of the vein-shaped biomimetic structure at different discharge rates is significantly lower than that of the other three structures. At a discharge rate of 5 C, the pressure drop (ΔP) is 31.416 Pa, and its overall heat dissipation performance is superior to that of the other structures. Further optimization showed that with a vein width of 6 mm, three coolant inlets, and an inlet flow rate of 0.3 m/s, Tmax of the system decreased to 300.9 K, temperature difference (ΔTmax) reached 2.31 K, and ΔP remained at 31.416 Pa.
  • Thermal simulation of a bidirectional counter-flow heat exchange plate for energy storage batteries
    WANG Bingbing, QIAO Jiafei, NA Ersu, WEN Xinyu, ZHUO Hua, ZHU Haitao, WANG Yongsheng
    Chinese Journal of Power Sources. 2025, 49(9): 1899-1906. https://doi.org/10.3969/j.issn.1002-087X.2025.09.015
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In response to the problem of high temperature and large temperature difference of energy storage batteries in the power grid under high-rate charging and discharging conditions, this study proposes a liquid cooling scheme based on bidirectional countercurrent heat exchange plates. The thermal management performance of three battery pack cooling schemes, namely bottom liquid cooling, side unidirectional flow cooling, and side bidirectional countercurrent cooling, was compared through numerical simulation under conventional charge discharge rate, high charge discharge rate, and ultra-high charge discharge rate conditions. The simulation results show that the traditional bottom liquid cooling scheme is limited by the heat transfer bottleneck in the top and bottom directions of the battery, making it difficult to meet the requirements of high-rate charging and discharging conditions. Although the side unidirectional cooling scheme improves heat transfer efficiency, it still cannot meet the control of temperature difference between the top and bottom of the battery during high-rate charging and discharging. The side bidirectional counter current heat exchange plate is designed with symmetrical flow channels to construct a reverse thermal compensation mechanism, successfully achieving dual control of maximum temperature difference of 4.8 K and maximum temperature of 299 K at high magnification. The bidirectional countercurrent heat exchange plate proposed in this study provides technical support for the safe operation of energy storage batteries in the power grid.
  • SOC prediction for lithium ion batteries based on the model of GA-FDFE-Seq2Seq
    ZHANG Nan, WU Kunzhen, LI Jing, LIU Yuqing, MA Yunfeng
    Chinese Journal of Power Sources. 2025, 49(9): 1907-1914. https://doi.org/10.3969/j.issn.1002-087X.2025.09.016
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In response to the diversity of state of charge (SOC) data features and the complexity of prediction, this study proposed a joint prediction model. This model integrates the Genetic Algorithm (GA), a Frequency-Domain Features Extractor (FDFE), and the Sequence to Sequence (Seq2Seq) network, which incorporates an attention mechanism. The FDFE extracts frequency information from lithium ion battery data and effectively analyzes the periodicity of data. Seq2Seq captures the temporal-frequency information from the data and the attention mechanism assigns weights, thereby the model establishes nonlinear and dynamic relationships between inputs and outputs. The GA optimizes hyper-parameters in the model, to allow the GA-FDFE-Seq2Seq model makes prediction under different conditions. The experimental results demonstrate that the GA-FDFE-Seq2Seq model has exhibited superior prediction performance with a prediction accuracy rate up to 99.99% and error below 1%. The GA-FDFE-Seq2Seq model boasts higher prediction accuracy, smaller errors, and stronger prediction reliability than other prediction models.
  • Estimation of lithium battery electro-thermal SOC based on dynamic capacity
    YANG Qiming, ZHAO Zining, ZHU Qiong, ZHOU Li, WANG Zeyuan
    Chinese Journal of Power Sources. 2025, 49(9): 1915-1925. https://doi.org/10.3969/j.issn.1002-087X.2025.09.017
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Accurately estimating the state of charge (SOC) of lithium batteries involves considering the real-time effects of temperature and discharge rate changes on battery capacity. Proposed an electro-thermal coupling method for state estimation based on dynamic capacity correction. An electro-thermal coupling model was built, factoring in the joint impact of SOC and temperature. By conducting hybrid pulse power characteristic and entropy heat coefficient experiments, electrical and thermal parameters were identified. The battery's capacity variation under various discharge rates and temperatures was tested to establish a correlation model among these factors. Then, using the corrected dynamic capacity, the H∞ filter was applied to the adaptive unscented Kalman filter (AUKF) algorithm to enable simultaneous online SOC and temperature estimation. Experimental validation under customised variable-power constant-current discharge conditions shows that this method boosts the accuracy and robustness of SOC and temperature estimation.
  • SOH estimation method for lithium iron phosphate batteries based on magnetic characterization
    WU Shang, HOU Chengwei, TANG Shun, CAO Yuancheng, OUYANG Zhongwen, WANG Zhenxing
    Chinese Journal of Power Sources. 2025, 49(9): 1926-1932. https://doi.org/10.3969/j.issn.1002-087X.2025.09.018
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    With the arrival of the retirement trend of lithium iron phosphate (LFP) batteries, the recycling and regeneration of waste LFP have become a recent research hotspot, among which the state of health (SOH) assessment of batteries is an important part of regeneration process. In order to provide a fast SOH evaluation method for commercial LFP batteries, constant current charge and discharge testing, XRD and magnetic characterization were used to study commercial LFP batteries after 1 cycle, 500 cycles, 1 000 cycles, and 3 000 cycles. The results of capacity measurement show that the SOH of the batteries are 100%, 90%, 37%, and 30% respectively. According to the magnetic characterization results, the SOH uncertainty of the batteries is evaluated to be no more than 7%. By examining the composition of the battery through magnetic characterization results, the original material is correlated with the loss of active lithium and SOH, providing a basis for the future development of in-situ non-destructive testing technology based on magnetic characterization.
  • Lithium ion battery reliability evaluation model considering state estimation
    RAO Bo, DU Jinqiao, TIAN Jie, LI Yan
    Chinese Journal of Power Sources. 2025, 49(9): 1933-1942. https://doi.org/10.3969/j.issn.1002-087X.2025.09.019
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Battery SOH is a key parameter for evaluating battery performance and predicting its remaining service life. In this paper, based on the second-order RC equivalent circuit model considering polarization and hysteresis effects, combined with the three-dimensional thermal model of the battery cell and the battery aging model based on capacity degradation and ohmic internal resistance, we constructed an electrically-thermally coupled model considering the effects of battery aging, and realized the estimation of battery SOC and SOH, and further got the multistate reliability assessment model of the energy storage battery. The results show that the proposed method can achieve higher accuracy of SOH estimation compared with the second-order RC equivalent circuit and the traditional electro-thermal coupling model algorithm, and the MAE and RMSE of the constructed electro-thermal coupling model taking into account the effect of battery aging are 0.181 and 0.259, which improves the estimation accuracy of the battery SOH compared with that of the traditional method, and further enhances the accuracy of the reliability assessment of the energy storage battery. The multistate reliability assessment model of energy storage battery provides the basis for the accurate calculation of energy storage battery reliability.
  • Remaining useful life prediction method for automotive lithium ion batteries based on MDA-PF under driving conditions
    LI Zhaojun, YANG Tongyu, ZHOU Yixin, WU Fangming, HUANG Wei
    Chinese Journal of Power Sources. 2025, 49(9): 1943-1950. https://doi.org/10.3969/j.issn.1002-087X.2025.09.020
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    A prediction method for remaining useful life (RUL) of lithium ion batteries based on a hybrid drive of data and models is proposed to address the complex capacity degradation characteristics and insufficient data of lithium ion batteries under driving conditions. Firstly, the Savitzky-Golay (SG) filtering method is applied to smooth and denoise the battery capacity degradation data. Then, establish a multi-source domain adaptation (MDA) neural network and use multiple sets of lithium ion battery capacity degradation data to predict the RUL of lithium ion batteries under low sample conditions. Subsequently, the particle filter (PF) algorithm was used to integrate the predicted values of the MDA neural network into the dynamic estimation process of the battery capacity decay empirical model, thereby forming the MDA-PF method that can achieve RUL prediction of lithium ion batteries under driving conditions. Finally, the proposed method is validated through examples. The experimental results show that the root mean square error of the prediction results using the proposed method is less than 0.13, the average absolute percentage error remains below 0.07, and the coefficient of determination is above 0.98. This indicates that the proposed MDA-PF method can effectively predict the RUL of lithium ion batteries for vehicles under driving conditions, and has better prediction performance compared to other commonly used methods.
  • Multitask prediction of lithium ion battery life based on convolutional neural network
    WANG Pengcheng, LAN Yuxiao, WU Changfeng, CAI Xiang, SUN Congcong, LU Guangbo, WANG Shiwen
    Chinese Journal of Power Sources. 2025, 49(9): 1951-1957. https://doi.org/10.3969/j.issn.1002-087X.2025.09.021
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Accurate prediction of the cycle life of lithium ion batteries is crucial for accelerating battery technology development and ensuring long-term reliable operation. However, the diversity of aging mechanisms, manufacturing and testing equipment variations, and differing operating conditions lead to inaccuracies in battery life prediction. Achieving precise battery life forecasting requires appropriate data characterization and effective prediction algorithms. This paper extracts voltage, current, and temperature data from initial charging cycles along with their variations across different cycles as input features to characterize battery status. Based on a multi-task learning framework, we employ a fused model integrating three-dimensional convolutional neural networks (3DCNN) and two-dimensional convolutional neural networks (2DCNN) to automatically extract features from input curves, explore relationships between different features and cycles, thereby predicting battery lifespan. Experimental results demonstrate that the proposed method achieves an early prediction error (after 20 cycles) of 5.01% across different batteries under various charging strategies.
  • Prediction of discharge capacity of lithium ion batteries based on random forest model
    ZHAO Jinnuo, WANG Yan, LI Yang
    Chinese Journal of Power Sources. 2025, 49(9): 1958-1961. https://doi.org/10.3969/j.issn.1002-087X.2025.09.022
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The discharge capacity is crucial for evaluating battery performance, and accurate prediction of discharge capacity is needed to optimize battery management and extend battery life. The decay rate of discharge capacity in lithium ion batteries has no linear pattern and is influenced by multiple factors, making it difficult to accurately predict. The BP neural network model widely used in such regression problems is prone to overfitting. To address this issue, this article constructs a discharge capacity prediction model based on the random forest algorithm, optimizes feature combinations using recursive feature elimination (RFE) method, solves the problem of model overfitting caused by high-dimensional data, and introduces K-fold cross validation to improve the model's generalization ability. This article uses 1 800 cycles of data from a certain energy storage lithium ion battery for testing, and compares and analyzes the prediction results with the BP neural network model to verify that the RF prediction model has better predictive performance than the BP neural network model. The experimental results show that the optimized RF model has a coefficient of determination (R2) of 0.97, a root mean square error (RMSE) of 0.062, and an average absolute error (MAE) of 0.28. It accurately characterizes the attenuation law of discharge capacity under different operating conditions and has better prediction performance than the BP neural network model, providing a reliable data-driven method for predicting the discharge capacity of lithium ion batteries.
Author Center
Review Center
Journal
Wechat
Copyright © Chinese Journal of Power Sources, All Rights Reserved.
Tel: 022-23959362　
E-mail：dianyuanjishu@cjpstj.com
Powered by Beijing Magtech Co. Ltd,.
    Total visitors:
    Visitors of today:
    Now online: