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.
20 December 2022 in IEEE Transactions on Industrial Electronics
Efficient battery capacity estimation is of utmost importance for safe and reliable operations of electric vehicles (EVs). This article proposes a battery capacity estimation framework based on real-world EV operating data collected from forty electric buses of the same model operating in two cities. First, a reference capacity calculation method is presented by combining the Coulomb counting method with the incremental capacity analysis method. Then, the impacts of temperature, current, and state-of-charge on battery degradation are quantitatively analyzed. Using the historical probability distributions as battery health features, a hybrid deep neural network model that combines a convolutional neural network with a fully connected neural network is proposed for battery capacity estimation. The validation results show that the proposed model outperforms the state-of-the-art methods and reaches a mean absolute percentage error of 2.79%, while maintaining low computational cost.
6 October 2023 in IEEE Transactions on Transportation Electrification
Accurately predicting the battery’s ageing trajectory is required to ensure the safe and reliable operation of electric vehicles (EVs), and is also the fundamental technique towards residual value assessment. As a critical enabler for mainstreaming EVs, fast charging has presented formidable challenges to health prognosis technology. This study systematically compares the performance of features extracted from the multi-step charging process in the state of health (SOH) assessment. First, twelve direct features are extracted from the voltage curve, and the degradation mechanisms strongly correlated to these features are analysed in detail. Integrating the degradation mechanism and correlation analysis, a data feature construction strategy is designed to categorise extracted features into groups. Then, the performance of different features extracted from the fast charging process in the SOH assessment is compared regarding estimation accuracy. Finally, the generalisation and feasibility of the optimal data feature are verified with different fast charging protocols and training data sizes. The verification results indicate that the data feature representing fused degradation modes has excellent generalisation and feasibility in SOH estimation, the mean absolute error (MAE) and root-mean-squared error (RMSE) for various cells under different decline patterns are within 0.90% and 1.10%, respectively.
The state-of-health (SOH) of lithium-ion batteries has a significant impact on the safety and reliability of electric vehicles. However, existing research on battery SOH estimation mainly relies on laboratory battery data and does not take into account the multi-faceted nature of battery aging, which limits the comprehensive and effective evaluation and prediction of battery health in real-world applications. To address these limitations, this study utilizes real electric vehicle operational data to propose a comprehensive battery health evaluation indicator and a deep learning predictive model. In this study, the battery capacity, ohmic resistance, and maximum output power were initially extracted as individual health indicators from actual vehicle operation data. Subsequently, a methodology that combines the improved criteria importance through inter-criteria correlation (CRITIC) weighting method with the grey relational analysis (GRA) method is employed to construct the comprehensive battery health evaluation indicator. Finally, a prediction model based on the attention mechanism and the bidirectional gated recurrent unit (Att-BiGRU) is proposed to forecast the comprehensive evaluation indicator. Experimental results using real-world vehicle data demonstrate that the proposed comprehensive health indicator can provide a thorough representation of the battery health state. Furthermore, the Att-BiGRU prediction model outperforms traditional machine learning models in terms of prediction accuracy.
Accurate and reliable health status prognostics are critical for ensuring battery safety and enabling smart management. Most existing studies assume stable operating conditions, which contrast with the dynamic and uncertain characteristics of in-service electric vehicles (EVs), thus challenging the practical application of developed approaches. To address this challenge, we propose an adaptable battery degradation prediction framework for EVs with different operating characteristics. Initially, we analyze the operational characteristics of EVs across different application scenarios and introduce a cluster-based charging pattern identification approach. Subsequently, we perform targeted feature extraction based on the identified charging patterns and propose a multilevel feature selection strategy to construct a comprehensive and effective feature pool. Furthermore, we develop two neural network (NN)-based models for reconstructing historical capacity trajectories and predicting battery degradation, further integrating transfer learning to enhance model efficiency and accuracy in unknown scenarios. Finally, we validate the proposed battery health prognostic framework across various training and prediction scenarios, demonstrating its high accuracy and reliability. Specifically, the mean absolute percentage error (MAPE) and root mean square error (RMSE) of degradation prediction are found to be within 1.30% and 2.05% for EVs in different operating scenarios, representing a notable improvement to existing methods.