Fault diagnosis for battery systems is essential for ensuring safe operation of electric vehicles (EVs). In this study, a novel model for battery fault diagnosis is established by combining the long short-term memory recurrent neural network (LSTM) and the equivalent circuit model (ECM). The modified adaptive boosting method is utilized to improve diagnosis accuracy, and a prejudging model is employed to reduce computational time and improve diagnosis reliability. Considering the influence of drivers’ behaviors on battery systems, the proposed diagnosis scheme is able to achieve potential failure risk assessment and accordingly to issue early thermal runaway warning. A large volume of real-world operation data is acquired from the National Monitoring and Management Center for New Energy Vehicles (NMMC-NEV) in China to verify its robustness, feasibility and reliability. The results show that the proposed method exhibits superior performance, and can achieve accurate fault prediction for potential battery cell failure and precise locating of thermal runaway cells.

Unmanaged cell inconsistency may cause accelerated battery degradation or even thermal runaway accidents in electric vehicles (EVs). Accurate cell inconsistency evaluation is a prerequisite for efficient battery health management to maintain safe and reliable operation and is also vital for battery second-life utilization. This article presents a cell inconsistency evaluation model for series-connected battery systems based on real-world EV operation data. The open-circuit voltage (OCV), internal resistance, and charging voltage curve are extracted as consistency indicators (CIs) from a large volume of electric taxis' operation data. The Thevenin equivalent circuit model is adopted to delineate battery dynamics, and an adaptive forgetting factor recursive least-squares method is proposed to reduce the fluctuation phenomenon in model parameter identification. With a modified robust regression method, the evolution characteristics of the three CIs are analyzed. The Mahalanobis distance in combination with the density-based spatial clustering of applications with noise is employed to comprehensively evaluate the multiparameter inconsistency state of a battery system based on the CIs. The results show that the proposed method can effectively assess cell inconsistency with high robustness and is competent for real-world applications.

The overdischarge can significantly degrade a lithium-ion (Li-ion) battery's lifetime. Therefore, it is important to detect the overdischarge and prevent severe damage of the Li-ion battery. Depending on the battery technology, there is a minimum voltage (cutoff voltage) that the battery is allowed to be discharged in common practice. Once the battery voltage is below the cutoff voltage, it is considered as overdischarge. However, overdischarge will not lead to immediate failure of the battery, and if it is not detected, the battery voltage can increase above the cutoff voltage during charging process. How to detect an overdischarge has happened, while the current voltage is larger than the cutoff voltage, thus becomes very challenging. In this article, a machine learning based two-layer overdischarge fault diagnosis strategy for Li-ion batteries in electric vehicles is proposed. The first layer is to detect the overdischarge by comparing the battery voltage with cutoff voltage, like what is utilized in common practice. If the battery voltage is larger than the cutoff voltage, the second layer, which is a detection approach based on eXtreme Gradient Boosting algorithm, is triggered. The second layer is employed to detect the previous overdischarge. The proposed method is validated by real electric vehicle data.

The incremental capacity analysis (ICA) method is a promising method in battery state of health (SOH) estimation studies. The incremental capacity (IC) curve determination is one of the critical parts of the ICA method. However, the uncertain and incomplete charging conditions of real-world electric vehicles (EVs) significantly limit the IC curve determination. This article provides a thorough analysis of the practicality and limitations of four IC curve determination methods based on the datasets collected from real-world operating EVs with a comprehensive comparison scheme. The Lorentz function fitting method is improved by the differential evolution algorithm, breaking the limitation of fixed parameter boundary constraints. A novel PCHIP method is further proposed to determine the IC curve, with higher robustness to realistic uncertain and incomplete charging conditions. The proposed method is validated by real-world data from 40 EVs with low sampling frequency. The results show the extracted features from the IC curves, determined by the proposed method have a stronger correlation with the SOH, allowing the accurate SOH estimation with a 2% error. With less computational resources and sampling frequency requirements, this method shows great potential for the realistic battery management system implementation.

Thermal runaway of an electric vehicle (EV) battery can cause severe loss of property and human life. With the increasing market share of EVs, this issue becomes more critical since one single cell short circuit could easily cause thermal runaway in a few minutes. Therefore, battery short-circuit detection systems are important for the prevention and limitation of EV fire incidents. This article proposes a short-circuit detection and isolation method for lithium-ion battery packs based on relative entropy and the Z-score method, which identifies the cell voltage dropping behaviors caused by a short circuit with the sliding window processing method. Taking the optimal sliding window width, the proposed detection algorithm can be performed in real time without any significant delay. The effectiveness and efficiency of the proposed method are verified using real-world data measured from EVs that experienced fire incidents caused by thermal runaways. Results indicate the method in this article is capable to recognize the data pattern of the potential threat in real time and send an early alarm to the driver.

Hundreds of thermal-runaway-induced battery fire accidents have been occurring to real-world electric vehicles (EVs) recent years, exposing life to danger and causing property losses. Timely and fast battery thermal runaway prognosis is essential but restricted by limited parameters and complex influencing factors during real-world operation of EVs, i.e., environment, driving behavior, and weather. To cope with the issue, several data-driven methods are combined and thermal runaway prognosis is realized by two steps, i.e., temperature prediction by the modified extreme gradient boosting (XGBoost) then abnormality detection by principal components analysis (PCA) and density-based spatial clustering of applications with noise (DBSCAN). The XGBoost is modified and trained by data of real-world EVs to couple the influencing factors during real-world operation of EVs. For parameter optimization, the ‘pre-training and adjacent grid optimizing method’(P-AGOM) and ’adjacent grid optimizing method’ (AGOM) are proposed to achieve locally optimal hyperparameters for XGBoost and DBSCAN. Verified results showcase the XGBoost-PCA-DBSCAN achieves accurate 5-minute-forward temperature prediction and the mean-square-errors (MSEs) of four seasons are only 0.0729, 0.0594, 0.0747, 0.0523 respectively. By modification of XGBoost, the MSE of temperature prediction is reduced by 31.2%. In addition, 35-minute-forward thermal runaway prognosis by the XGBoost-PCA-DBSCAN will provide driver sufficient response time to minimize the loss of life and property.