Battery Fault Research Articles

Early Warning of Energy Storage Battery Fault Based on Improved Autoformer and Adaptive Threshold

To enhance voltage prediction accuracy in energy storage batteries and address the limitations of fixed threshold warning methods, a fault warning approach based on an improved Autoformer model and adaptive thresholds is proposed. First, a spatiotemporal filtering layer is introduced into the autocorrelation mechanism to analyze the trend features of voltage sequences across different frequency domains. Additionally, an adaptive gating residual connection is used to link the sublayer and current layer output features, which helps to improve the model's adaptive feature selection capability. This innovation enables the development of a robust voltage prediction model based on the enhanced Autoformer. Then, a similarity‐based adaptive threshold, using interval estimation, is employed to rapidly track variations in battery voltage, enabling dynamic adjustment of voltage thresholds. Finally, the proposed method is validated with real voltage data from an operational energy storage station. The experimental results shows that the proposed model has higher accuracy and robustness compared to similar methods. The adaptive threshold can reduce the false alarm rate by ≈18% and issue alarms at three sampling points ahead of the battery management system alarm, improving fault warning accuracy and illustrating that early fault warning is effectively and practically carried out using the method.

Series arc-induced internal short circuit leading to thermal runaway in lithium-ion battery

With the widespread implementation of battery energy storage systems (BESSs), significant attention has been focused on issues involving electrical safety. The series arc hazard caused by loose connectors between batteries has become a serious problem. However, research findings for the evolution process of the series arc and the related hazard principle are still unclear. Therefore, in this study we focus on the series arc at the negative terminal of a 20 Ah prismatic lithium-ion battery, establish an experimental platform for the arc, and conduct research on the hazards process. Our results indicate that the arc can induce the thermal failure of the battery when the power supply voltage is 300 V and the circuit current is 15 A. Through a battery voltage analysis, computed tomography scans, and jellyroll disassembly, we uncover the evolution process and hazard laws of series arcs and clarify the failure pathways of arc-induced battery faults. The hotspots formed by arc melt the casing and cause electrolyte leakage. In addition, the heat transfer from the battery terminal to the jellyroll induces separator melting and internal short circuits in batteries. These cause an internal short circuit between the anode and the cathode, as well as combustion of the leaked electrolyte, which give rise to distinct thermal runaway behavior under different states of charge. By comparing runaway behavior with failures triggered by external heating, we clarify that the series arc is a novel risk factor that induces failure. This study addresses the gap in research related to arc effects on battery safety. This is crucial to the development of safe battery systems that do not present arc hazards.

MgTe for efficient adsorption of lithium-ion battery fault gases (H2, CO, and CO2) and failure risk index calculation: A DFT computational study

In recent years, lithium-ion batteries have been used in bulk for the development of new energy vehicles. Studying fault gas adsorption in lithium-ion batteries is crucial to their safe operation. This study constructs and optimizes the adsorption model for fault gases adsorbed on the MgTe monolayer. Analysis of the energy- and electron-related adsorption properties indicates that the MgTe monolayer has a strong adsorption effect on lithium-ion battery failure gases. Meanwhile, the value of Eads increases as the gas coverage expands, with CO showing the greatest increase at 144.79%. Secondly, the conductivity change response of CO adsorption on MgTe is obvious, and the Desorption Time is considerable, indicating its potential as a long-life adsorbent. Furthermore, the Risk Index is defined by the change of Eads (CO), and a strong correlation between the Risk Index and lithium-ion batteries is obtained. This study provides a new idea for lithium-ion battery failure warning and is of great value for the widespread application of new energy.

Intelligent fault diagnosis for unbalanced battery data using adversarial domain expansion and enhanced stochastic configuration networks

An accurate and efficient fault diagnosis method for battery systems is crucial to ensuring the safety of battery packs. Addressing the issue of insufficient actual fault data in battery operations, this paper proposes an intelligent fault diagnosis method based on feature-enhanced stochastic configuration networks and adversarial domain expansion of imbalanced battery fault data (AFDEM-FESCN). Firstly, we designed an adversarial fault domain data expansion method (AFDEM). By learning the distribution of fault data through adversarial training, the distribution of sample domains is balanced, thereby reducing model bias. Subsequently, we adjusted the distribution of SCN iterative parameters and added a linear feature layer. This enhances the feature extraction capability of the network through distribution overlay, enabling fault diagnosis. Finally, the effectiveness and feasibility of the proposed method were validated through a practical battery system fault diagnosis case, achieving a diagnostic accuracy of 92.1%. Experimental results demonstrate that the AFDEM-FESCN method exhibits good accuracy in battery system fault diagnosis, providing an effective solution to the challenge of imbalanced data in intelligent fault diagnosis.

Improvement of local outlier factor algorithms for lithium-ion battery fault diagnosis

Fault diagnosis is one of the most important active strategies to protect lithium-ion batteries (LIBs) from safety accidents. The tasks of fault diagnosis usually can be divided into three levels, i.e., (1) fault detection, (2) fault isolation, and (3) fault estimation. The local outlier factor (LOF) method has been proved effective in conducting fault detection (level 1 of fault diagnosis) for LIB energy storage systems (ESSs). However, the original LOF algorithm (OLOF) may fail to make correct fault estimation (level 3 of fault diagnosis). Therefore, in this study, two novel LOF algorithms, i.e., the improved LOF algorithm (ILOF) and the simplified LOF algorithm (SLOF), are put forward to enhance the capability of fault estimation. The performance of the three LOF algorithms: OLOF, ILOF and SLOF, are firstly compared via three mathematical cases. Their capabilities in fault diagnosis are further investigated based on the simulated data from an air-cooled LIB ESS and the experimental data from a water-cooled LIB ESS. Results prove the effectiveness of all the three algorithms in fault detection; nevertheless, when the OLOF algorithm fails to make correct fault estimation, both of the two proposed novel algorithms, i.e., the ILOF algorithm and the SLOF algorithm, present the capability in making correct fault estimation. It is found as well that the ILOF algorithm is better than the SLOF algorithm in robustness though the latter is more efficient than the former in memory usage and computational time.

Fault diagnosis method for lithium-ion batteries based on the combination of voltage prediction and Z-score

ABSTRACT Safety accidents in new energy electric vehicles caused by lithium-ion battery failures occur frequently, and the timely and accurate diagnosis of failures in battery packs is crucial. Voltage, as one of the primary characterization parameters of lithium-ion battery malfunctions, is widely utilized in fault diagnosis. This article proposes a lithium-ion battery fault diagnosis method Fault diagnosis method based on the combination of voltage prediction and Z-score. Firstly, the stable trend component is extracted from the battery voltage data using variational mode decomposition, which avoids the influence of noisy signals and random perturbations to the greatest extent. Subsequently, a TCN-BiLSTM-attention model is designed to estimate the average voltage of the battery under normal conditions. Finally, the residuals between the estimated and individual cell voltages are calculated, and the Z-score is utilized to locate and judge whether the battery is caused by the occurrence of a fault. Through verification with real vehicle data and experimental data, the proposed method effectively identifies abnormal battery cells. Compared to the correlation coefficient method, this approach exhibits superior applicability.

A Remote Intelligent Check and Discharge Test System for Substation Battery Pack Based on Raspberry Pi

Abstract Regular check and discharge test of the battery pack is an effective means to ensure the normal state of the battery and timely detect outdated batteries. However, conventional manual check and discharge test methods are costly, cumbersome, and inefficient in maintenance. This article introduces a remote intelligent check and discharge test system for a substation battery pack based on Raspberry Pi. The system monitors the battery status in real-time, allowing operators to detect battery faults on time through the front-end management interface. At the same time, remote check and discharge tests can be carried out, greatly ensuring battery safety and solving the shortcomings of conventional manual methods.

An early-fault diagnostic method based on phase plane for lithium-ion batteries under complex operation conditions

Timely and accurate diagnosis of lithium-ion battery faults is crucial for ensuring the safety of electric vehicles. While incremental capacity (IC)-based diagnostic methods are effective, they can only be employed under charging conditions. To address this issue, a fault diagnostic method based on phase plane analysis is proposed to extend the application to various operating conditions. In a phase plane, the battery voltage is plotted on the x-axis, while the differential in voltage is plotted on the y-axis. During charging, diagnostic vectors indicating the occurrence and type of faults are extracted from the trajectories in the phase plane. Under complex operating conditions, early faults can be detected based on the morphological and scale deviation of the trajectories. Experimental results have shown that the proposed method can effectively detect minor faults. The fault detection rate and detection accuracy rate under charging conditions are 96.25 % and 97.73 %, respectively, which are 17.5 % and 8.54 % higher than those of the IC-based diagnostic method. Furthermore, the fault detection rate and accuracy rate under UDDS (Urban Dynamometer Driving Schedule) are 91.76 % and 97.23 %, respectively. Additionally, the diagnostic speed of the proposed method is twice as fast as that of the IC-based method.

Early Diagnosis of Battery Faults Through an Unsupervised Health Scoring Method for Real-World Applications

Early Diagnosis of Battery Faults Through an Unsupervised Health Scoring Method for Real-World Applications

Review of Lithium-Ion Battery Fault Features, Diagnosis Methods, and Diagnosis Procedures

Review of Lithium-Ion Battery Fault Features, Diagnosis Methods, and Diagnosis Procedures

Adaptive fault detection for lithium-ion battery combining physical model-based observer and BiLSTMNN learning approach

Lithium-ion batteries possess one of the best energy-weight ratios, while safety and reliability are critical issues for its continued operation. Due to extreme fast-charging, huge application sizes and variable usage environment, the risk of soft battery faults, including soft internal short circuit (SISC), is increased and may evolve into severe accidents. Therefore, accurate early detection of lithium-ion battery fault is imperative to guarantee the battery performance. Motivated by this fact, we proposed a real time fault detection framework for battery soft faults. Based on the Equivalent Circuit Model (ECM) and coupling thermal model, Extended Kalman Filter (EKF) observer is used for reliable monitoring of battery voltage and surface temperature. Inspired by uncertainty interference during the observation process, we explore an observed-based learning approach that combine physical model-based observer and Bidirectional Long Short-Term Memory neural networks (BiLSTMNN) to empower the observer with the ability to learn the uncertainties, which is efficient in distinguishing soft faults information and uncertainties from residual. Furthermore, the memory characteristics of BiLSTMNN are considered to improve the robustness of the detection system, including comprehensive training data and optimized input features. Finally, through the comparative analysis and the robustness evaluation experiments including various driving cycles and internal short circuits (ISC) faults test verity the good performance of the proposed method.

A novel battery abnormality diagnosis method using multi-scale normalized coefficient of variation in real-world vehicles

Accurate and efficient diagnosis of battery voltage abnormality is crucial for the safe operation of electric vehicles. This paper proposes an innovative battery voltage abnormality diagnosis method based on a normalized coefficient of variation in real-world electric vehicles. Vehicle and laboratory data are collected and analyzed, with joint preprocessing to improve data quality, and battery voltages are log-transformed to improve the contribution of anomalous voltage fluctuations. The normalized coefficient of variation is proposed to detect the fluctuation inconsistency of cell voltage, and the risk coefficient rule is formulated by Z-score and normalization. Furthermore, the validity and robustness are verified by laboratory and real-world battery faults. The results demonstrate that the optimal slide step and calculation window for real-world under-voltage fault are 10 and 40, and those for laboratory lithium plating and real-world thermal runaway are both 10 and 50, respectively. More importantly, this study introduces a battery abnormality diagnosis strategy based on the vehicle T-box, anticipated to be widely implemented to ensure the safety of real-vehicle operations. This method not only enhances the accuracy and efficiency of detecting electric vehicle battery abnormalities, but also offers a practical solution to prevent battery related faults.

Transformer-based lithium battery fault diagnosis research for electric vehicles

With the fast advances of new energy vehicles, the EV battery technology needs to be further improved to follow the step. How to effectively diagnose the electric vehicle’s lithium battery fault becomes a hotspot in the academic circle. This study has proposed new method that uses the state of charge of the battery and self-coder depth to detect faults in the lithium battery group of electric vehicles. First, the study investigates the single lithium battery faults. Then, it builds a lithium battery group fault diagnosis model by integrating the battery charge state and denoising converter network. Finally, it uses a dataset and retired battery group to validate the model’s performance. The results show that the proposed model achieves an accuracy of 93.18% and a recall rate of 93.73% in identifying the faults in the lithium batteries of the electric vehicles and its F1 value is as high as 0.95. Moreover, the modeling method has the lowest prediction error, indicating its high accuracy and robustness in diagnosing the faults of battery packs. This study aims to provide an effective solution for the fault diagnosis of lithium battery packs in electric vehicles.

Fault Diagnosis Method for Lithium-Ion Power Battery Incorporating Multidimensional Fault Features

Accurately identifying a specific faulty monomer in a battery pack in the early stages of battery failure is essential to preventing safety accidents and minimizing property damage. While there are existing lithium-ion power battery fault diagnosis methods used in laboratory settings, their effectiveness in real-world vehicle conditions is limited. To address this, fault diagnosis methods for real-vehicle conditions should incorporate fault characteristic parameters based on external battery fault characterization, enabling the accurate identification of different fault types. However, these methods are constrained when confronted with complex fault types. To overcome these limitations, this paper proposes a battery fault diagnosis method that combines multidimensional fault features. By merging different fault feature parameters and mapping them to a high-dimensional space, the method utilizes a local outlier factor (LOF) algorithm to detect anomalous values, enabling fault diagnosis in complex working conditions. This method improves the detection time by an average of 22 min compared to the extended RMSE method and maintains strong robustness while correctly detecting faults compared to other conventional methods.

Fault diagnosis of lithium-ion batteries based on wavelet packet decomposition and Manhattan average distance

ABSTRACT As lithium-ion batteries are widely used in electric vehicles, safety accidents caused by battery failures emerge one after another. Nevertheless, failures caused by changes in the internal structure or characteristics of the battery, such as sudden and progressive failures, are still a serious problem for electric vehicles, challenging existing fault diagnosis methods. This paper first performs wavelet packet decomposition on the battery’s raw voltage signal to obtain high-quality low-frequency and high-frequency characteristic signal components. Then performs singular value decomposition on the characteristic signal components to extract the corresponding singular value characteristic parameters, and introduces the Manhattan average distance algorithm to battery faults. Diagnosing and locating faulty battery units using the Laida criterion (3-σ criterion) outlier detection method. Finally, actual vehicle data were used to verify the reliability, stability, accuracy of the method, and compared with the traditional Manhattan distance, correlation coefficient, information entropy methods. The method in this paper has good fault detection effects on vehicles with sudden and progressive faults vehicles.

Fault Diagnosis and Detection for Battery System in Real-World Electric Vehicles Based on Long-Term Feature Outlier Analysis

Accurate detection and diagnosis battery faults are increasingly important to guarantee safety and reliability of battery systems. Developed methods for battery early fault diagnosis concentrate on short-term data to analyze the deviation of external features without considering the long-term latent period of faults. This work proposes a novel data-driven method to detect long-term latent fault and abnormality for electric vehicles based on real-world operation data. Specifically, the battery fault features are extracted from the incremental capacity curves, which are smoothed by advanced filter algorithms. Secondly, principal component analysis algorithm is utilized to reduce dimensionality and the cumulative percent variance is to determine the number of significant features. Based on the features, a cluster algorithm is employed to capture the battery potential failure information. Moreover, the cumulative root-mean-square-deviation is introduced to quantificationally analyze the degree of the battery failures using large-scale battery data to avoid the missing fault reports using short-term data. In addition, a battery system failure index is proposed to evaluate battery fault conditions. The results indicate that the proposed long-term feature analysis method can effectively detect and diagnose faults.

Internal short circuit fault diagnosis for the lithium-ion batteries with unknown parameters based on transfer learning optimized residual network by multi-label data processing

In recent years, internal short circuits have frequently been the cause of safety accidents, and traditional diagnosis methods for such problem have limitations that prevent their application under complex working conditions. Furthermore, due to the scarcity of real internal short circuit data, it is difficult to obtain a large amount of real vehicle data, which makes internal short circuit fault diagnosis even more challenging. In this paper, fault batteries with unknown parameters are diagnosed using a residual network based on multi-label processed battery data and transfer learning is utilized to optimize the diagnosis effect for fault localization and fault degree identification, thus increasing the information richness of the training model samples and improving the training effect. Accuracy, recall, and false positive rate are used to evaluate the diagnosis results of the residual network. The results demonstrate that the accuracy rate increased by 11.1% after transfer learning, indicating a significant reduction in misdiagnoses. The proposed fault diagnosis method using residual network and transfer learning has practical significance for enhancing battery fault detection accuracy and efficiency. Especially in areas such as electric vehicles and energy storage systems, it can provide effective technical support for battery health monitoring and early warning.

Multi-fault detection and diagnosis method for battery packs based on statistical analysis

Rapid and accurate battery fault diagnosis and distinction is of great importance in electrical vehicles and electrochemical energy storage system. However, misdiagnosis and missed diagnosis happened occasionally. In this paper, a statistical analysis-based multi-fault diagnosis method is proposed to detect and localize short circuit faults, electrical connection faults and voltage sensor faults in LFP battery packs. This method uses non-redundant interleaved voltage measurement topology to detect battery voltages, where every voltage sensor measures the sum of two neighboring batteries and one connection resistor between them. The statistical analysis method sets detection thresholds based on the battery operating data, and captures fault characteristics by analyzing abnormal changes in battery voltage unrelated to current. Theoretical analysis and tests verified that this method can diagnose these three kinds of faults. Sensor faults of excessive error and data sticking can also be distinguished.

The Early Detection of Faults for Lithium-Ion Batteries in Energy Storage Systems Using Independent Component Analysis with Mahalanobis Distance

In recent years, battery fires have become more common owing to the increased use of lithium-ion batteries. Therefore, monitoring technology is required to detect battery anomalies because battery fires cause significant damage to systems. We used Mahalanobis distance (MD) and independent component analysis (ICA) to detect early battery faults in a real-world energy storage system (ESS). The fault types included historical data of battery overvoltage and humidity anomaly alarms generated by the system management program. These are typical preliminary symptoms of thermal runaway, the leading cause of lithium-ion battery fires. The alarms were generated by the system management program based on thresholds. If a fire occurs in an ESS, the humidity inside the ESS will increase very quickly, which means that threshold-based alarm generation methods can be risky. In addition, industrial datasets contain many outliers for various reasons, including measurement and communication errors in sensors. These outliers can lead to biased training results for models. Therefore, we used MD to remove outliers and performed fault detection based on ICA. The proposed method determines confidence limits based on statistics derived from normal samples with outliers removed, resulting in well-defined thresholds compared to existing fault detection methods. Moreover, it demonstrated the ability to detect faults earlier than the point at which alarms were generated by the system management program: 15 min earlier for battery overvoltage and 26 min earlier for humidity anomalies.

Optimizing fault diagnosis for electric vehicle battery systems: Improved Giza pyramids construction and advanced gradient boosting decision trees

This manuscript proposes an online multi-fault diagnostic technique for the series string batteries in electric vehicles to identify and diagnose external/internal short circuits, sensor faults, and connection fault detection. The proposed technique combines the Improved Giza Pyramids Construction and Gradient Boosting Decision Trees. Many variables, including road conditions, electromagnetic interference, and driving habits, may complicate the battery fault system, multi-parameter or non-linear coupling, making it challenging to diagnose faults properly by using a single parameter in the real operation of an electric vehicle. The voltage fluctuation data is collected by simulating a battery discharging and charging investigation in a vibration setting. By deconstructing and rebuilding the discrete wavelet transform, the proposed approach removes voltage signal noise. To categorize the fault state, the voltage change, co-variance and variance matrixes, and voltage characteristics are employed as input-values in the improved Giza pyramids construction, and gradient boosting decision trees technique. The Improved Giza pyramids construction technique is utilized to detect fault signals and determine the degree of the defect. The cell faults are separated from other faults using the Gradient Boosting Decision Trees approach by identifying the surrounding voltages with fault flags. Furthermore, the correlation coefficient of the nearby voltage difference and current isolates connection and voltage sensor defects. The multi-fault diagnosis approach may prevent erroneous fault detection and offer high resilience to battery inconsistencies and normal measurement errors of state of health, ambient temperature, and state of charge. The projected control theme is evaluated on the MATLAB platform and its performance is assessed. The performances of the proposed method are compared with existing techniques like Artificial Bee Colony, Differential Evolution, Improved Giza Pyramids Construction and Particle Swarm Optimization. The efficiency of the Artificial Bee Colony, Differential Evolution, Particle Swarm Optimization, and Improved Giza Pyramids Construction and proposed technique is 82.237 %, 79.265 %, 87.1029 %, 89.023 % and 95.4501 %.