Fei Feng

The safety of lithium-ion batteries is a critical bottleneck restricting their further development in the fields of new energy vehicles and energy storage, with safety incidents caused by thermal runaway being particularly severe. To achieve accurate prediction of the electro-thermal behavior of lithium-ion batteries under normal and extreme operating conditions, this study proposes an electrochemical-thermal coupled modeling method based on a Physics-Informed Neural Network (PINN). This method first establishes an electrochemical-thermal coupled mechanistic model that combines the Single Particle Model (SPM) with a lumped thermal model. To address the challenge of sparse experimental data on thermal runaway, we used COMSOL Multiphysics to generate simulation data under multiple operating conditions and combined it with partial high-fidelity experimental data to form a hybrid training dataset. Based on this, a dual-channel PINN architecture consisting of an electrochemical network and a thermal network was designed, embedding the governing partial differential equations of the coupled model into the neural network’s loss function as a physical constraint, thereby achieving a deep fusion of model-driven and data-driven approaches. Finally, the model’s performance was validated through conventional charge-discharge experiments at different C-rates and heating-induced thermal runaway experiments at various states of charge (SOC). The results show that the proposed model can predict the battery’s voltage and temperature responses with high accuracy under both normal charge-discharge and extreme thermal runaway conditions, demonstrating excellent robustness and high fidelity.

Lithium-ion batteries (LIBs) are a cornerstone technology driving transportation electrification and renewable energy storage systems

Lithium-ion batteries pose a risk of thermal runaway (TR) during operation. Monitoring the temperature distribution within the battery is crucial for ensuring thermal safety and preventing such runaway events. However, constructing the temperature field typically requires a large amount of sensor data and is challenging to perform in real time. To address this issue, this article proposes a novel method for reconstructing the temperature field of lithium-ion batteries during TR using a sparse random-sensor mask autoencoder (SRS-MAE). This approach enables real-time reconstruction of the battery’s complete pixel-level temperature field using minimal sensor data. Additionally, this article explores the impact of sensor location and quantity on the model’s reconstruction accuracy, providing a theoretical basis for optimizing the position of temperature sensors in actual battery management systems. This article employs an experimental-simulated thermal image dataset to validate the method’s superiority and reliability. The results demonstrate that the SRS-MAE can accurately predict the battery’s TR temperature field, requiring data from as few as two sensors to reconstruct the complete pixel-level temperature field of the battery. Given the critical need for efficient and accurate temperature monitoring in battery systems, the SRS-MAE method offers a vital solution for battery temperature field reconstruction and advancing the capabilities of battery management systems.

A fusion of local and global feature-based explainable diagnosis methods is proposed for lithium-ion battery (LIB) degradation modes in this article. The method captures fault features from the discharge curves of a semi-battery model. Voltage curves are transformed into incremental capacity (IC) curves, converted into 2-D images and subjected to feature extraction to obtain high-dimensional features. A battery fault feature-extraction module based on a local-global feature fusion module is designed to enhance feature saliency by extracting long-term dependency information. To explain the learning process, a battery fault explainability determination module is designed to analyze the attention points of the proposed method on the overall IC curve image changes by highlighting the focus areas. The explainability analysis of the decision-making process was accomplished by integrating the IC curve analysis, including variations in the IC peaks, to establish a physical correlation. Finally, the effectiveness and superiority of the proposed method were verified through experiments.

The remaining useful life (RUL) prediction is one of the most important tasks in the prognostics and health management of industrial equipment. The statistical model-based method is widely used for RUL prediction, but it depends on sufficient prior knowledge and appropriate degradation assumptions, which limits its use in the case of the complexity degradation process or insufficient prior knowledge. Motivated by this, we propose a hybrid model to describe the degradation process, which is composed of some given degradation models. This can be conducive to describing complex degradation trajectories and further improving the flexibility of the degradation model. Then, a sparsity mechanism is proposed to automatically fuse these candidate degradation models based on the sparse variational Bayesian method. As a result, we can find the most appropriate degradation mechanism for a given degradation process without sufficient prior knowledge. To the best of the authors' knowledge, it is the first time to use such a fusion mechanism to describe the complex degradation process. A numerical example, a practical example, and three public datasets are used to verify the effectiveness and merits of the proposed method.

To guarantee the safety of electric vehicles (EVs), abusive operations should be strictly prohibited for EV-mounted li-ion batteries (LiBs). This article proposes an intelligent strategy to infer the historical abusive operations (HAOs) on LiBs by retrospectively analyzing the relationship between HAO-induced damages and electrical behaviors. First, the electrical synchronicity among the peer LiB cells in a series module is perceived using an improved correlation coefficient formula; then the synchronicity sequences are translated into recurrence plot images (RCPIs) and Gramian angular field images (GAFIs) that can provide a wealth of textures regarding cross-time autocorrelations. Second, based on the hierarchical clustering algorithm, a pilot test is performed to preliminarily examine the separability of these images corresponding to different HAOs. Finally, the GoogLeNet model is employed to model the causality between image features and HAO specifics, whereby the type and intensity of potential HAO can be inferred. A realistic dataset is obtained by inflicting abuses such as overvoltage, overheat, and vibration on LiB cells. Experimental verifications show that the proposed strategy performs well in backtracking the HAOs on LiBs by providing effective and reliable inferences on HAO specifics. The accuracy rates of HAO type assessment and severity evaluation can achieve about 77% and 76% using RCPIs, and about 79% and 76% using GAFIs, respectively.

Timely and accurate fault diagnosis for a lithium-ion battery pack is critical to ensure its safety. However, the early fault of a battery pack is difficult to detect because of its unobvious fault effect and nonlinear time-varying characteristics. In this paper, a fault diagnosis method based on relative entropy and state of charge (SOC) estimation is proposed to detect fault in lithium-ion batteries. First, the relative entropies of the voltage, temperature and SOC of battery cells are calculated by using a sliding window, and the cumulative sum (CUSUM) test is adopted to achieve fault diagnosis and isolation. Second, the SOC estimation of the short-circuit cell is obtained, and the short-circuit resistance is estimated for a quantitative analysis of the short-circuit fault. Furthermore, the effectiveness of our method is validated by multiple fault tests in a thermally coupled electrochemical battery model. The results show that the proposed method can accurately detect different types of faults and evaluate the short-circuit fault degree by resistance estimation. The voltage/temperature sensor fault is detected at 71 s/58 s after faults have occurred, and a short-circuit fault is diagnosed at 111 s after the fault. In addition, the standard error deviation of short-circuit resistance estimation is less than 0.12 Ω/0.33 Ω for a 5 Ω/10 Ω short-circuit resistor.

To address the challenges of efficient state monitoring of lithium-ion batteries in electric vehicles, a co-estimation algorithm of state-of-charge (SOC) and state-of-health (SOH) is developed. The algorithm integrates techniques of adaptive recursive least squares and dual adaptive extended Kalman filtering to enhance robustness, mitigate data saturation, and reduce the impact of colored noise. At 25 °C, the algorithm is tested and verified under dynamic stress test (DST) and Beijing bus DST conditions. Under the Beijing bus DST condition, the algorithm achieves a mean absolute error (MAE) of 0.17% and a root mean square error (RMSE) of 0.19% for SOC estimation, with a convergence time of 4 s. Under the DST condition, the corresponding values are 0.05% for MAE, 0.07% for RMSE, and 5 s for convergence time. Moreover, in this research, the SOH is described as having internal resistance. Under the Beijing bus DST condition, the MAE and the RMSE of the estimated internal resistance of the proposed approach are 0.018% and 0.075%, with the corresponding values of 0.014% and 0.043% under the DST condition. The results of the experiments provide empirical evidence for the challenges associated with the efficacious estimation of SOC and SOH.