Search

Summary: The battery pack of electric vehicles (EV) is generally composed of multiple cells in series. Due to the inconsistency between the cells in the production process and use stage, the capacity and state‐of‐charge (SOC) of the cells will be different. We propose an online estimation method based on the charging curve similarity principle in this paper. The proposed method uses a series of charging time differences (CTD) during the charge. By analyzing the CTD curve, the capacity and SOC difference can be achieved. The first‐order resistance circuit model is used for the series charging curve simulation. Further experimental verification is conducted using two groups of four cells in series. In simulations and experiments, the error of the proposed capacity estimation method and the initial SOC error are less than 1%. Finally, the robustness of the proposed method is verified using EV cloud data. The results demonstrate that the proposed method has good robustness at the level of EV cloud data. [ABSTRACT FROM AUTHOR]

Internal short circuit is considered as one of the general causes that may lead to battery thermal runaway. The capacity of cells ages with the effect of working conditions. Hence, both micro-short circuit (MSC) and low-capacity cells may exist in a battery pack. However, both two faults perform the same features in the discharging process: state of charge (SOC) deviation increases continuously. If we diagnosed the abnormal states only based on the discharging data, it would misdiagnose these two faults. A fault identification method based on capacity estimation is proposed to distinguish MSC and low-capacity cells in the article. Furthermore, the capacity of the low-capacity cell and the MSC cell could be quantitatively estimated. In the proposed method, a mean-difference model and extended Kalman filter algorithm are used to calculate the cell SOCs in the battery pack. An online capacity estimation method is adopted to estimate the cell capacities in the charging and discharging process. A reasonable threshold considering capacity change characteristics is established to initially identify the fault and for further quantitative diagnosis. The experimental results show that a coexisting MSC fault and low-capacity fault in the battery packs could be diagnosed effectively by using the proposed method. [ABSTRACT FROM AUTHOR]

During the usage of electric vehicles, the battery decays and the cell variations expand in the battery pack. In the discharge process, both the low-capacity cell and the micro-short-circuit (MSC) cell have the abnormal feature that the state-of-charge (SOC) differences increase continuously. Hence, a low-capacity cell is likely to be misdiagnosed as an MSC cell, and vice versa. In this article, a fault identification approach based on mutual information is proposed to detect the MSC cell and low-capacity cell. A decision tree for fault identification is established by analyzing the battery fault characteristics of the short circuit, low capacity, and the abnormality of initial SOC difference. It is pointed out that the SOC deviation of the low-capacity cell is related to the mean SOC, while that of the MSC cell is related to time. A low-pass filter is used to get internal resistance differences in order to achieve the SOC deviations based on the cell different model. Finally, the MSC cell and the low-capacity cell can be identified using the mutual information which can quantitatively calculate the correlation between the SOC deviation and the mean SOC. Experimental results prove that the proposed method is reliable to identify the MSC cell and the low-capacity cell. [ABSTRACT FROM AUTHOR]

Summary: Lithium‐ion battery accidents occur frequently, and thermal runaway accidents are common. Lithium plating may occur in batteries at different temperatures, state of charge (SOCs), charging rates, etc., and lithium plating at the anode is one of the important incentives for thermal runaway of batteries. The plating of lithium metal is closely related to the anode potential of the battery. In this paper, based on the anode potential sensor, the anode potential of the battery under various working conditions is detected in real‐time, and the direct anode equivalent circuit model (DAECM) and the indirect anode equivalent circuit model (IDAECM) are proposed. The DAECM directly estimates the anode potential based on identifying anode parameters. Therefore, the IDAECM is proposed to estimate the cathode potential directly based on cathode parameter identification, to estimate the anode potential indirectly. First, the high‐precision P2D model is used to simulate the cathode, anode, and full cell potentials under various working conditions. According to the 1/50C discharge rate test, the open circuit voltage (OCV)‐SOC curve of the cell is obtained. The NEDC conditions are used to identify the parameters of the model, and the values of each parameter of the two models under different SOCs are obtained. According to the parameter identification results above, the anode potential under multirate constant current charging (MRCC) test conditions is estimated. Based on the method above, the IDAECM model is suitable for estimating the anode potential in the early stage of fast charging with large rate constant current, while the DAECM model is suitable for the final stage of fast charging with small rate constant current. [ABSTRACT FROM AUTHOR]

Summary: With the commercialization of the electric vehicles, the large‐scale lithium‐ion cells as the power of electric vehicles are to be retired. The second‐use of retired cells is of great significance to improve the battery economy. A fast classification and regroup evaluation method of the retired lithium‐ion cells are proposed in this paper to improve the classification efficiency of retired lithium‐ion cells and adapt to the regroup under different conditions. The lithium‐ion cells after being balanced in parallel are charged in series with a constant current. A support vector regression (SVR) model with the parameters optimized by the particle swarm optimization (PSO) algorithm is built for the fast capacity estimation and the error will not exceed 0.3%. Different cells regrouped means different performance. In order to improve the consistency of retired cells and satisfy different using conditions, a Weighted‐K‐means algorithm is proposed in this paper to regroup the cells with the known capacity and internal resistance. The classification method is evaluated by the voltage consistency of cells using different working conditions, which indicates capacity occupied a large proportion can meet the requirement of energy condition meanwhile keep a good consistency. But the resistance will dominate in algorithm under conditions which have requirement for instantaneous power. [ABSTRACT FROM AUTHOR]

Summary: With the commercialization of the electric vehicles, the large‐scale lithium‐ion cells as the power of electric vehicles are to be retired. The second‐use of retired cells is of great significance to improve the battery economy. A fast classification and regroup evaluation method of the retired lithium‐ion cells are proposed in this paper to improve the classification efficiency of retired lithium‐ion cells and adapt to the regroup under different conditions. The lithium‐ion cells after being balanced in parallel are charged in series with a constant current. A support vector regression (SVR) model with the parameters optimized by the particle swarm optimization (PSO) algorithm is built for the fast capacity estimation and the error will not exceed 0.3%. Different cells regrouped means different performance. In order to improve the consistency of retired cells and satisfy different using conditions, a Weighted‐K‐means algorithm is proposed in this paper to regroup the cells with the known capacity and internal resistance. The classification method is evaluated by the voltage consistency of cells using different working conditions, which indicates capacity occupied a large proportion can meet the requirement of energy condition meanwhile keep a good consistency. But the resistance will dominate in algorithm under conditions which have requirement for instantaneous power. [ABSTRACT FROM AUTHOR]

Real-time battery capacity estimation is very important for the battery management but usually has a low accuracy in electric vehicles due to the complicated real working conditions and the changing parameters during the battery lifespan. Traditional estimation methods, e.g. methods based on the empirical models such as the Arrhenius capacity aging model, or methods based on the state of charge, always suffer from the parameters mismatch during the long battery lifespan. In this paper, we put forward a method based on charging curve sections which can be easily achieved for electric vehicles. The proposed method uses the complete charging curves and the corresponding capacities in experiments as the training data for a certain battery type. The optimal fixed voltage window is then determined by the particle swarm optimization with a designed objective function focused on minimizing the error of linear capacity loss assumption. The capacity is finally estimated by calculating the charging capacities during the optimal fixed voltage window online. The proposed method is verified using the designed experimental data, and the error is proved to be small. • A method based on charging curve sections is proposed to estimate battery capacity. • Optimal fixed voltage window for linear capacity loss estimation is found. • Long-time and various aging experiments are carried out for verification. • Proposed method has a good accuracy during the battery lifespan [ABSTRACT FROM AUTHOR]

Micro-short-circuit (MSC) is a latent risk in power batteries, which may give rise to thermal runaway and even catastrophic safety hazards. The motivation of this paper is to quantitatively analyze MSC in an initial stage, particularly for lithium-ion batteries. To verify the feasibility of the proposed method, an equivalent MSC experiment is carried out. Based on a cell difference model, the cell state of charge (SOC) differences with the mean SOC for a battery pack are estimated by extended Kalman filter. The evaluated SOC difference can track the actual value well. Furthermore, an MSC diagnostic method is developed by employing recursive least squares filter. The method is demonstrated to examine the short-circuit resistance accurately. The results also show that the proposed method requires low computational load for the SOC difference and short-circuit resistance diagnosis. [ABSTRACT FROM AUTHOR]

Lithium-ion cells connected in series are prone to be overdischarged, which may induce internal short circuit (ISC). In this study, the electrical behavior of overdischarge-induced ISC in NMC cells is analyzed. The performance of 11 new cells is tested, and these cells are then overdischarged to different depths. The electrical characteristics of these overdischarged cells are examined to assess the relationship between the ISC and depth of discharge (DOD). The cells are then rested for 100 days to assess the self-repair capability of overdischarged cells. The ISC is characterized by comparing the DOD with the capacity loss, leakage current, and ISC resistance. The degree of the ISC is found to increase nonlinearly with the DOD. A DOD of 120% or more can induce an irreversible ISC, and the electrical characteristics of severely overdischarged cells are similar to linear resistance characteristics. The self-repair behavior occurs in the cells with a DOD of less than 120% after a longtime rest and leads to a decrease in the degree of the ISC. The cyclic test results show that a higher DOD leads to a higher capacity decay rate. The electrical characteristics of the overdischarge-induced ISC are valuable in ISC identification and prediction. [ABSTRACT FROM AUTHOR]

State-of-charge (SOC) inconsistency impacts the power, durability and safety of the battery pack. Therefore, it is necessary to measure the SOC inconsistency of the battery pack with good accuracy. We explore a novel method for modeling and estimating the SOC inconsistency of lithium-ion (Li-ion) battery pack with low computation effort. In this method, a second-order RC model is selected as the cell mean model (CMM) to represent the overall performance of the battery pack. A hypothetical Rint model is employed as the cell difference model (CDM) to evaluate the SOC difference. The parameters of mean-difference model (MDM) are identified with particle swarm optimization (PSO). Subsequently, the mean SOC and the cell SOC differences are estimated by using extended Kalman filter (EKF). Finally, we conduct an experiment on a small Li-ion battery pack with twelve cells connected in series. The results show that the evaluated SOC difference is capable of tracking the changing of actual value after a quick convergence. [ABSTRACT FROM AUTHOR]

Sate of charge (SOC) estimation is generally acknowledged as one of the most important functions in battery management system for lithium-ion batteries in new energy vehicles. Though every effort is made for various online SOC estimation methods to reliably increase the estimation accuracy as much as possible within the limited on-chip resources, little literature discusses the error sources for those SOC estimation methods. This paper firstly reviews the commonly studied SOC estimation methods from a conventional classification. A novel perspective focusing on the error analysis of the SOC estimation methods is proposed. SOC estimation methods are analyzed from the views of the measured values, models, algorithms and state parameters. Subsequently, the error flow charts are proposed to analyze the error sources from the signal measurement to the models and algorithms for the widely used online SOC estimation methods in new energy vehicles. Finally, with the consideration of the working conditions, choosing more reliable and applicable SOC estimation methods is discussed, and the future development of the promising online SOC estimation methods is suggested. [ABSTRACT FROM AUTHOR]

An appropriate model is a prerequisite for accurate state-of-charge (SOC) estimation. The widely used equivalent circuit models (ECMs) employ a variety of forms; thus, to find the optimum ECM is a primary task for SOC estimation. In this work, we examined eleven ECMs to fulfill the following goals: (1) to compare the typical ECMs for accuracy, stability, and robustness of model and SOC estimation; (2) to compare and evaluate the robustness of the ECMs considering model and sensor errors. The results indicate that the model accuracy does not always improve by increasing the order of the RC network. Conversely, over-fitting problems appear with a certain probability. The first- and second-order RC models are the best choice owing to their balance of accuracy and reliability for LiNMC batteries. The higher-order RC model has better robustness considering the variation in model parameters and sensor errors. Independently of the ECM adopted, an accurate OCV-SOC curve and high precision sensors are essential. [ABSTRACT FROM AUTHOR]

As one single cell cannot meet power and driving range requirement in an electric vehicle, the battery packs with hundreds of single cells connected in parallel and series should be constructed. The most significant difference between a single cell and a battery pack is cell variation. Not only does cell variation affect pack energy density and power density, but also it causes early degradation of battery and potential safety issues. The cell variation effects on battery packs are studied, which are of great significant to battery pack screening and management scheme. In this study, the description for the consistency characteristics of battery packs was first proposed and a pack model with 96 cells connected in series was established. A set of parameters are introduced to study the cell variation and their impacts on battery packs are analyzed through the battery pack capacity loss simulation and experiments. Meanwhile, the capacity loss composition of the battery pack is obtained and verified by the temperature variation experiment. The results from this research can demonstrate that the temperature, self-discharge rate and coulombic efficiency are the major affecting parameters of cell variation and indicate the dissipative cell equalization is sufficient for the battery pack. [ABSTRACT FROM AUTHOR]

One of the key battery performances which can be described as the resistance characteristic is the cell voltage response with dynamic currents, and it requires synchronous cell voltages and current. But due to inevitable network latency in modular battery management system (BMS) with controller area network (CAN), cell voltages and current are usually asynchronous. We firstly analyze the sampling and the storage process of battery signals to study the asynchronous mechanism in BMS. We develop an on-line synchronization method using a “global clock” from the master controller to decrease the time delay as much as possible. And we further propose a model based sync method based on the frequency division equivalent circuit model (FDECM) for the battery pack. The low frequency cell difference model is used to identify cell “resistances difference”, and then the optimal time compensation for cell voltages is obtained when the minimum mean absolute derivative (MAD) value of identified resistance differences is reached according to the low frequency characteristic of cell “resistances difference”. The proposed methods are verified by simulation and experiment. The current and cell voltages in the data logger of BMS can be synchronized when the optimal compensation time is applied respectively for each cell. The data after synchronization can meet the requirements of further data analysis and processing, which is of great significance to enhance and improve the control strategy of BMS. [ABSTRACT FROM AUTHOR]

Massive data storage is an advanced function in a fully functional battery management system (BMS). Reducing the recording signal length undoubtedly saves the precious memory space for BMS. And it also reduces the network and computation loads. However, it leads to a side effect that the trend of signal distortion is enhanced. The optimal recording frequency in practice should be as low as possible on the condition that little signal distortion happens. This paper presents a novel method which uses a multi-frequency recording technology that cooperates two approaches according to the signal dynamics. A flexible recording frequency method is applied for stationary signals which only records signals when their values are changed. While for dynamic signals, the most dynamic period is found using discrete wavelet transformation (DWT) and further analyzed by fast Fourier transformation (FFT). By comparing two recording signal indicators for four different recording frequencies, we conclude that recording at 1 Hz is not qualified for the cell voltage and current during the dynamic period in our system due to the high dynamic performance of the vehicle. In the demonstrated vehicle, only by increasing the recording frequency to at least 2 Hz, can the accuracy of the recorded cell voltage achieve the level the same as the measurement accuracy in engineering. And we also verify that when the recording frequency is reduced to the optimal frequency compared to the high frequency recorded original signals, the accuracy of the SOC estimation is not influenced. [ABSTRACT FROM AUTHOR]

Coulombic efficiency (CE) is an important parameter for battery cells. Unfortunately, precise measurement of CE is extremely difficult, so CE is seldom focused during cell performance study. Nevertheless, correlation between stage of charge (SOC) and CE has significant impact on battery pack performance, while the influences of different correlations between SOC and CE on battery packs have never been explored as far as we know. We present a novel method which implements two series-connected cells that can determine the correlation between SOC and CE. The experimental result shows that CE is almost invariant with SOC changes for the experimental commercial LiFePO 4 /C cells. This paper further investigates series-connected cells with three typical correlations between SOC and CE in simulations. The simulation results indicate that cells with negative correlations between CE and SOC are preferred for series-connected battery packs, as they tend to diminish SOC difference and increase the pack capacity which can be considered as a self-balancing mechanism. [ABSTRACT FROM AUTHOR]

Battery cell capacity loss is extensively studied so as to extend battery life in varied applications from portable consumer electronics to energy storage devices. Battery packs are constructed especially in energy storage devices to provide sufficient voltage and capacity. However, engineering practice indicates that battery packs always fade more critically than cells. We investigate the evolution of battery pack capacity loss by analyzing cell aging mechanisms using the “Electric quantity – Capacity Scatter Diagram (ECSD)” from a system point of view. The results show that cell capacity loss is not the sole contributor to pack capacity loss. The loss of lithium inventory variation at anodes between cells plays a significant role in pack capacity evolution. Therefore, we suggest more attention could be paid to the loss of lithium inventory at anodes in order to mitigate pack capacity degradation. [ABSTRACT FROM AUTHOR]

Micro short circuit (MSC) is a potential risk of thermal runaway in batteries. It is essential to prevent thermal runaway and improve the safety of batteries via short-circuit detection. Traditional detection methods take the healthy cells in the battery pack as a reference, which use statistical characteristics to perform qualitative or quantitative MSC diagnosis. However, for the application scenarios of one single cell, the existing methods cannot determine short circuit due to the lack of healthy batteries as a reference. Therefore, a quantitative diagnosis method for single lithium-ion cell applications is proposed in this paper. The core idea of the method is that the estimated capacity of the short-circuit cell during the discharging process is smaller than the normal value, while the estimated capacity during the charging process is larger than the normal value. Hence, by comparing the historical capacity variation characteristics under the charging and discharging cycle, the fault can be diagnosed quantitatively. The experimental results show when the short-circuit resistance is 5Ω for large-capacity cells, the short-circuit resistance estimation accuracy can reach 2.5%. And the capacity estimation error of the short-circuit cell is within 1.5% after the capacity is compensated. • The method can realize the quantitative short-circuit diagnosis. • The method is suitable for single lithium-ion cell applications. • It can be diagnosed through the longitudinal comparison of historical data. • It is based on the comparison of the historical charge and discharge capacity. • It can improve the accuracy of fault diagnosis. [ABSTRACT FROM AUTHOR]

Abstract: Because of the inevitable inconsistency during manufacture and use of battery cells, cell variations in battery packs have significant impacts on battery pack capacities, durability and safety for electric vehicles (EVs). To reduce cell variations and increase pack capacity, cell equalization is essentially required. In the series of two papers, we discover that dissipative cell equalization (DCE) using dissipative resistances is a feasible on-line equalization method for battery packs in EVs. We subsequently propose on-line equalization algorithms for lithium-ion battery packs based on charging cell voltage curves (CCVCs). The objective of these algorithms is to maximize pack capacities by conditioning CCVCs. As the first paper of the series, we first briefly review equalization topologies and algorithms. We discover cell remaining charging capacity (RCC) can be on-line estimated and further propose DCE algorithm based on remaining charging capacity estimation (RCCE). We establish a pack model with 8 cells in series and simulate 4 scenes with different cell variations. RCCE–DCE algorithm is proved to be effective by comparing pack capacities with/without RCCE–DCE algorithm. The equalization capability and over-equalization prevention are further examined, and the result shows that RCCE–DCE algorithm is suitable for on-line equalization in EVs. [Copyright &y& Elsevier]

Abstract: In the first part of this work, we propose dissipative cell equalization (DCE) algorithm based on remaining charging capacity estimation (RCCE) and establish a pack model with 8 cells in series. The results show that RCCE-DCE algorithm is suitable for on-line equalization in electric vehicles (EVs) and no over-equalization happens. However, 1% pack capacity difference from the DCE theoretical pack capacity is observed with RCCE-DCE algorithm. Therefore, as the second part of the series, we propose fuzzy logic (FL) DCE algorithm based on charging cell voltage curves (CCVCs). Cell capacities and SOCs are fuzzily identified in FL-DCE algorithm by comparing cell voltages at the beginning and end of charging. Adaptive FL-DCE is further improved to prevent over-equalization and maintain the equalization capability. The simulation results show that pack capacity difference from the DCE theoretical pack capacity with the adaptive FL-DCE is smaller than that with RCCE-DCE algorithm, and the duration of the infant stage is also shorter. The proposed adaptive FL-DCE is suitable for on-line equalization in EVs and well prevents over-equalization. [Copyright &y& Elsevier]

Highlights: [•] We propose a Mean-plus-Difference Model (M+D Model) for battery pack. [•] Mean model represents the whole pack and difference model investigates inconsistency. [•] OCV differences are estimated using cell difference model for LiFePO4 pack. [•] SOC strategy is proposed to identify SOC inconsistency by lowering pack SOC in HEV. [•] SOC differences are determined using SOC-difference/OCV-difference curve. [Copyright &y& Elsevier]

The empirical model is an open-loop model for battery capacity prediction, which faces the general problem of parameter mismatch. The data-driven method can realize the closed-loop estimation of capacity, but the data quality and model adaptability may lead to large noise. This paper proposes a framework of battery capacity prediction based on the feedforward empirical model and the feedback data-driven method. With the difference of the estimated capacities from the two models, parameters of the empirical model are modified, and the fusion capacity is finally predicted. Two cases based on different empirical models and fusion methods are introduced, one is based on discrete Arrhenius aging model with sequent extended Kalman filters (Case Ⅰ)，and the other is based on the linear aging model with incremental PID + Luenberger observer (Case Ⅱ). The two cases are verified by the aging experimental data. The results show that under the proposed capacity prediction framework, the model parameters can be effectively corrected and gradually converged, and the accurate capacity prediction can be achieved with battery aging. With the low calculation load, the predicted fusion capacity can still guarantee high accuracy in Case Ⅱ, which is a good choice for the online capacity prediction. • Fusion capacity prediction based on empirical model and data-driven method. • The parameters of empirical model are modified by the parameters feedback correction algorithm. • The fusion capacity prediction is achieved by the capacity fusion correction algorithm. • Two cases are introduced to illustrate the applicability of the method framework. [ABSTRACT FROM AUTHOR]

Abstract: Because of the diversiform driving conditions and the cell variations, it is difficult to accurately determine battery pack capacities in electric vehicles (EVs) by model prediction or direct measurement. This paper studies the charging cell voltage curves (CCVC) for the estimation of the LiFePO4 battery pack capacities in EVs. We propose the uniform CCVC hypothesis and estimate cell capacities by overlapping CCVCs using CCVC transformation. CCVCs of two LiFePO4 cells with large capacity difference are used to verify the hypothesis. We further develop an equivalent simplified approach using voltage-capacity rate curve (VCRC) and implement Genetic Algorithm (GA) to find the optimum transformation parameter for overlapping VCRCs. A small battery pack with four LiFePO4 cells in series is employed to verify the method and the result shows that the estimation errors of both pack capacity and cell capacities are less than 1%. With the proposed method, the battery pack capacity can be precisely estimated which could be used for the driving range prediction. Meanwhile, the estimated cell capacities in battery packs will significantly support the study of cell degradation and cell variations in vehicle driving conditions. [Copyright &y& Elsevier]

Abstract: Fault diagnosis for inconsistent cells in a battery pack is important for electric vehicle (EV) safety. Discovering the fault and identifying its cause on-site are extremely challenging. The literature has rarely explored such issues. This paper presents a battery pack system in a demonstrated EV with 96 cells in series and discovers the battery power fade fault during the demonstration. The preliminary analysis, after collecting data and preprocessing the typical data periods, shows that the internal or contact resistance increase causes the fault. To further identify the fault cause, we introduce an equivalent circuit model (ECM) and identify the cell resistances using the total least squares algorithm. A diagnostic method is proposed, using the identified cell resistances from 3 months demonstration data. Calculating the Shannon entropy clearly identifies the cause of the power fade fault. Appropriate measures are taken to solve the fault, and the latent safety issue is eliminated. The proposed diagnostic method can be used on-site for EV diagnosis without laboratory testing methods. [Copyright &y& Elsevier]

Lithium plating can cause capacity fade and thermal runaway safety issues in lithium-ion batteries. Therefore, accurately detecting the amount of lithium plating on the surface of the battery's negative electrode is crucial for battery safety. This is especially crucial in high-energy-density applications such as battery energy storage systems or in electric vehicles (EVs). Early detection of lithium plating is crucial for evaluation of reliability and longevity. It also serves as a method for early diagnostics in practical industrial applications or infrastructure, such as EV transportation. This can enhance its impact on customers. This study validates the effectiveness of titration gas chromatography (TGC) technology in quantitatively detecting lithium plating on graphite negative electrodes in lithium-ion batteries. The results show that it can detect a minimum of 2.4 μmol of metallic lithium. Compared with the heating direct current resistance and reference electrode methods, which can be used to perform only qualitative dendrite detection, TGC has a wider range of detection. Compared with the nuclear magnetic resonance (NMR) method with higher quantitative detection accuracy, the maximum difference between the detection results of the two methods was only 7.2%, but the TGC method had lower cost and higher implementation convenience. In summary, among various dendrite detection methods, the TGC method can not only realize the effective quantitative detection of lithium plating, but also comprehensively consider its detection range, implementation convenience, cost, and detection accuracy, indicating that it is suitable for engineering applications and has the prospect of realizing large-scale quantitative detection of lithium plating in lithium-ion batteries. [ABSTRACT FROM AUTHOR]

The traditional time-domain equivalent circuit model (ECM) for Li-ion batteries may achieve high accuracy by identifying model parameters through limited dynamic conditions. However, the traditional ECM (TECM) parameter matching techniques based on time-domain may have the problem of adaptability of working conditions. To solve the problem of dependence on time-domain working conditions and improve the stability of the full-cycle modeling of lithium batteries, it is critical to establish the ECM based on the principle of electrochemical impedance spectroscopy (EIS). However, the full-frequency domain ECM structures (FECM) and the corresponding parameters based on the whole frequency ranges of the EIS employ too many resources for on-line estimation. In this paper, we firstly analyze the frequency region of the standard operating conditions of electric vehicles and theoretically discuss the rationality of the Simplified ECM base on the low-frequency region of the EIS. The proposed low-frequency domain ECM (LECM) simplifies the elements that describe the medium-high frequency regions of the EIS. It only uses the low-frequency region of the EIS to identify the model parameters. The results show that the accuracy of the proposed LECM is almost the same as that of FECM, and saves an average of 40% of the calculation load. • Impedance-based battery model employing one ZARC-element. • The lithium-ion battery operates at low-frequency domains in electric vehicles. • Analyze the sensitivity of EIS in different frequency domains. • Validation of the proposed approach under three dynamic working conditions. • Accurate voltage estimation within an RSME range of 5 mV. [ABSTRACT FROM AUTHOR]

Packs with high self-discharge accelerate the capacity decline and even cause the safe issues. It is important to keep the self-discharge rate at a uniform and small level for all the cells in a pack. The traditional clustering methods are either costly or time-consuming. In this paper, a novel classification method is invented to realize the fast estimation of the relative self-discharge rate. Firstly, the balancing technology for large-scale cells is proposed to ensure the voltage equalization. Small batches of equalized cells are then clamped in detection circuits to realize the external mapping of the internal current originated from the self-discharge. The relative self-discharge rates of the cells are estimated by the equivalent parallel circuit model with self-discharge. The results show that the method can detect the cell with a relative self-discharge rate of 3%/month for three cells at a time within 1 h. To further classify the cells according to the relative self-discharge rate, microammeters are connected in each branch of the detection circuits. The results show that the method can identify the relative self-discharge rate in less than 2 h, and the errors can be kept within 0.1%/month. • A fast and economic relative self-discharge rate detection method is proposed. • The relative self-discharge rate is estimated by monitoring branch currents. • The method can estimate relative self-discharge rate and realize classification. • The method is verified by experiment and simulation with good accuracy. • The method is fast within 2 h. [ABSTRACT FROM AUTHOR]

Summary: Battery inconsistency in electric vehicles is an important factor causing battery capacity degradation and safety problems. Therefore, battery equalization technology plays an important role in improving the performance and safety of battery packs. Among the existing equalization technologies, passive equalization is inefficient and active equalization is expensive and complex. Most of the active equalization algorithms only consider the balance of the cells within the battery module, but the balance between modules are not involved. However, battery packs for electric vehicles often consist of multiple modules, cooperative equalization between modules are essentially required to improve the balance efficiency. A novel cooperative equalization system for multi‐modules in the battery pack is proposed in this paper. The system combines active and passive equalization, and also includes a fast discharge function for balancing modules by a power resistor. An equalization algorithm aiming at the optimal equalization time is studied. Finally, the equalization system is verified in simulation and experiment. The experimental results show that the proposed equalizer demonstrates good performance in different consistency cases using the proposed control strategy and optimization equalization algorithm. [ABSTRACT FROM AUTHOR]

• An online capacity estimation method based on charging curve sections is proposed. • A discrete Arrhenius aging model is applied for the continuous capacity estimation. • Parameters of the discrete Arrhenius aging model are updated by the first EKF. • The second EKF makes a fusion capacity estimation of the aforementioned methods. • The fusion capacity estimation is accurate over the life span of the battery cell. Practical open-loop capacity estimation models, such as the Arrhenius aging model, require parameter updating for the real-world capacity estimation in order to guarantee the estimation accuracy. In this paper, a novel capacity estimation method for lithium-ion batteries, based on the fusion estimation of charging curve sections and the discrete Arrhenius aging model using sequential extended Kalman filters, is proposed. The estimation method based on fractional charging curves is developed to estimate the battery capacity during vehicle charging, and the estimation results serve as the feedback using the first Kalman filter to update the model parameters of the discrete Arrhenius aging model. Then, the second Kalman filter makes a fusion capacity estimation based on the results of charging curve sections and the discrete Arrhenius aging model with the modified parameters. The results of the cycle life tests show that the proposed algorithm can modify the parameters of the discrete Arrhenius aging model online. And the fusion capacity estimation error is less than 1% when the model parameters reach a steady state. [ABSTRACT FROM AUTHOR]

The service life of large battery packs can be significantly influenced by only one or two abnormal cells with faster aging rates. However, the early‐stage identification of lifetime abnormality is challenging due to the low abnormal rate and imperceptible initial performance deviations. This work proposes a lifetime abnormality detection method for batteries based on few‐shot learning and using only the first‐cycle aging data. Verified with the largest known dataset with 215 commercial lithium‐ion batteries, the method can identify all abnormal batteries, with a false alarm rate of only 3.8%. It is also found that any capacity and resistance‐based approach can easily fail to screen out a large proportion of the abnormal batteries, which should be given enough attention. This work highlights the opportunities to diagnose lifetime abnormalities via "big data" analysis, without requiring additional experimental effort or battery sensors, thereby leading to extended battery life, increased cost‐benefit, and improved environmental friendliness. [ABSTRACT FROM AUTHOR]

The remaining discharge energy (RDE) estimation of lithium-ion batteries heavily depends on the battery's future working conditions. However, the traditional time series-based method for predicting future working conditions is too burdensome to be applied online. In this study, an RDE estimation method based on average working condition prediction and multi-parameter updating is proposed. First, the ohmic resistance of batteries is identified online, the temperature-aging factor is introduced against battery aging and temperature changes, and the OCV-SOC is estimated by curve scaling. Then, the future working conditions of the battery are predicted based on the average working condition prediction with less calculation. Finally, the RDE is estimated under complex working conditions. The experimental results show that the RDE estimation error of the battery is less than 3% during battery aging and temperature changes under complex working conditions. Moreover, in addition, the computational burden of the proposed method is only 1% of that of traditional methods, making it very suitable for online applications. [ABSTRACT FROM AUTHOR]

Summary: The peak power of a battery is a vital feature for electric vehicles to maximize battery efficiency and ensure the safe operation of the system. Currently, the estimation and prediction of the state‐of‐power are based either on precise model algorithms or a large amount of test data. However, these methods will lead to conservative measurements. In this study, 11 Ah LiFePO4 cells are selected as the research object, and a new peak power test experimental procedure is proposed, which is mainly based on the constant power charge/discharge modes. Moreover, the conception of power sensitivity is proposed, which will make the procedure easier to determine the set power during the constant power experiments. The results indicate that the method can assist researchers in achieving a full understanding of the peak power characteristic of batteries. [ABSTRACT FROM AUTHOR]

Summary: Lithium‐ion pouch cells are widely used in electric vehicles because of their high energy density than other structures. There is an unavoidable external surface pressure between the cells in the process of packing and driving of electric vehicles. The influence of external surface pressure on the main properties of the lithium‐ion pouch cell has been studied, which is of great significance to packing batteries and reusing retired cells. In this study, a testing device applied for the measurement of constant external surface pressures of lithium‐ion pouch cells was first proposed and the different pressure stress‐strain distribution on the external surface of cells under semirigid material pads were analyzed by simulation. The effects of pressure on the capacity, internal resistance, and open circuit voltage of fresh and aged LiNixCoyMnzO2 (NCM) lithium‐ion pouch cells are analyzed through experiments under 1 Mpa external surface pressure. The results show that the internal resistance of fresh cells tends to decrease. The average percentage of the maximum reduction of internal resistance is 13.28%. The experiments also demonstrate that the capacity of aging cells increased by 2.3%. Irreversibility of capacity improvement indicates applying appropriate external surface pressure can improve the secondary utilization efficiency of aging cells. [ABSTRACT FROM AUTHOR]

Abstract Retired cells from electric vehicles could provide considerable economic benefits through secondary uses such as energy storage. However, the screening and regrouping of large-scale cells is facing the problem of low efficiency and low accuracy. To address this problem, two rapid and accurate screening approaches are proposed in this study. Firstly, the characteristics of the series-charging curve of large-scale retired cells are investigated. Then, two novel screening models, namely the neural network model and the piecewise linear fitting model, are thus constructed using the capacity and voltage profiles of a small number of sample cells, and the capacity of a large number of cells can be estimated in batches. Moreover, a device for fast switching between series and parallel is designed to improve efficiency, and a regrouping approach for different second-use applications is introduced. Finally, the proposed approaches are verified by simulations and experiments. The main results are as follow: (1) The piecewise linear fitting model should be used in the small sample case, and the neural network model should be adopted in the large sample case for higher estimation accuracy; (2) The proposed approaches are feasible and effective, and several cases illustrate that the capacity estimation error of the proposed approaches is less than 4%; (3) The screening efficiency of the proposed approaches increases with the increase in the number of cells, and screening results of 5000 cells indicate that the screening efficiency of our proposed approach is at least 5 times higher than that of a traditional approach. Highlights • Two rapid and accurate screening approaches for the secondary applications of retired cells are proposed. • Simulation and experimental results show that the proposed approaches are feasible and effective. • A device for fast-switching between series and parallel is designed to improve screening efficiency. • A regrouping approach based on screening results for different second-use applications is introduced. [ABSTRACT FROM AUTHOR]

Abstract A suitable model structure and matched model parameters are prerequisites for the precise estimation of the battery states. Previous studies pay little attention to whether a parameter identification method is suitable for a model. In this study, a comparative study is conducted by implementing model parameter optimization for nine equivalent circuit models using nine optimizers in the entire SOC area. The following conclusions are drawn: (1) PNGV and the exact algorithms are an ideal combination in the low SOC area (0%–20%). (2) In the high SOC area (20–100%), exact algorithms are an ideal choice for the first-order RC models, and PSO is an ideal identification algorithm for second-order RC models. For the third- and fourth-order RC models, firefly algorithm has the highest accuracy with longer identification time. (3) Firefly algorithm has the superior capacity to identify the accurate model parameters and PSO has the best comprehensive performance for on-line parameter identification. Highlights • A comparative study on model parameter identification for nine models using nine optimizers in the entire SOC area. • PNGV and exact algorithms are an ideal combination in the low SOC area. • In the high SOC area, EAs and PSO are the ideal choice for the first- and second-order RC models, respectively. • For the third- and fourth-order RC models, firefly algorithm has the highest accuracy with longer identification time. • FA has excellent identification accuracy and PSO has the best comprehensive performance for on-line identification. [ABSTRACT FROM AUTHOR]

Abstract The fast diagnosis of micro-short circuit cells is crucial for the safety of battery packs. Based on the difference between the "median cell" and other cells in a battery pack, we propose a method that can identify micro-short circuit cells under dynamic conditions in real time. We model cell differences and analyze the model from the perspective of its low frequency variation characteristics. We find that approximate open-circuit voltage differences can be obtained when terminal voltage differences are passed through low-pass filters. Then approximate electric quantity differences can be obtained by utilizing the open-circuit voltage differences and the data smoothing function of low-pass filters. For onboard applications of diagnosis method, the recursive least square is adopted to estimate micro-short circuit currents and resistances utilizing the change of electric quantity differences. We verify and analyze the feasibility of the diagnosis method by using simulation data when the cells in a battery pack have temperature, state of charge, capacity, and internal resistance inconsistency, respectively. Finally, the effectiveness of the diagnosis method is further verified by the triggering experiments of micro-short circuits for real battery packs. Highlights • A real-time diagnosis method of micro-short circuit (MSC) fault is proposed. • Diagnosis method is analyzed and validated in case of cell inconsistency. • The concept of median cell is proposed and applied to battery fault diagnosis. • Application of recursive least squares (RLS) method realizes online estimation of MSC current. [ABSTRACT FROM AUTHOR]

Micro-short circuit (MSC) of a lithium-ion battery cell is a potential safety hazard for battery packs. How to identify the cell with MSC in the latent phase before a thermal runaway becomes a difficult problem to solve. We propose a diagnosis method to detect the MSC according to the remaining charging capacity (RCC) variations between cells. When the charging cell voltage curve (CCVC) of the first fully charged cell is regarded as the benchmark, the RCC of each cell can be obtained by the CCVC transformation based on the uniform CCVC hypothesis. The accuracy of the RCC estimation method is validated under constant current and constant power charging experiments. The leakage current of the MSC cell can be obtained by the increase of the RCC after each charge, and it can be converted into the MSC resistance. We set up the battery pack system model with the MSC fault in Simulink ® to verify the effectiveness of the method when the battery pack is charged up to different charge cutoff voltages. Through a series of experiments with external resistors, the adaptability of the MSC diagnosis method is further validated for the aged cell, multi-stage charging, and constant power charging. [ABSTRACT FROM AUTHOR]

To improve the energy density of lithium-ion battery packs, lithium-ion batteries are gradually advancing towards large-size structures, which has become one of the dominant development trends in the battery industry. With large-size blade lithium-ion batteries as the research object, this paper develops a high-precision electro-thermal coupling model based on the relevant parameters obtained through basic performance experiments, explores the mechanism of battery inhomogeneity from a simulation perspective, and further proposes a design management method. First of all, the optimal intervals of capacity and temperature, as well as the characteristics of the inhomogeneity distribution for large-size cells, are determined by essential performance and inhomogeneity tests; subsequently, the electrochemical and thermal characteristics of the large-size battery are described precisely through a 3D thermoelectric coupling mechanism model, and the inhomogeneity of the temperature distribution is obtained through simulation; eventually, the optimized cell connection method and thermal management strategy are proposed based on the validated model. As indicated by the findings, the above solutions effectively ease the inhomogeneity of large-size cells and significantly boost the performance of large-size cells under different operating conditions. [ABSTRACT FROM AUTHOR]

State-of-charge (SOC) estimation of lithium-ion batteries (LIBs) is the basis of other state estimations. However, its accuracy can be affected by many factors, such as temperature and ageing. To handle this bottleneck issue, we here propose a joint SOC-SOH estimation method considering the influence of the temperature. It combines the Forgetting Factor Recursive Least Squares (FFRLS) algorithm, Total Least Squares (TLS) algorithm, and Unscented Kalman Filter (UKF) algorithm. First, the FFRLS algorithm is used to identify and update the parameters of the equivalent circuit model in real time under different battery ageing degrees. Then, the TLS algorithm is used to estimate the battery SOH to improve the prior estimation accuracy of SOC. Next, the SOC is calculated by the UKF algorithm, and finally, a more accurate SOH can be obtained according to the UKF-based SOC trajectory. The battery-in-the-loop experiments are utilized to verify the proposed algorithm. For the cases of temperature change up to 35 °C and capacity decay up to 10%, our joint estimator can achieve ultra-low errors, bounded by 2%, respectively, for SOH and SOC. The proposed method paves the way for the advancement of battery use in applications, such as electric vehicles and microgrid applications. [ABSTRACT FROM AUTHOR]

Under fast-charging or low-temperature charging conditions, large-format lithium-ion batteries are prone to inhomogeneous lithium plating, resulting in localized degradation and even internal short circuit, thus requiring a thorough investigation. This study establishes a three-dimensional electrochemical model to analyze the inhomogeneous lithium plating behavior in lithium-ion batteries. The model is validated with the voltage profiles and temperature distributions of a 24Ah pouch lithium-ion battery during low-temperature charging. The inhomogeneous lithium plating induced by non-uniform temperature distribution and the underlying mechanism are investigated using the model. The local current density and overpotential turn out to be the key factors affecting the spatial distribution of plated lithium. Moreover, the effects of temperature gradients on the lithium plating and stripping behaviors are explored. In-plane temperature gradient shows little effect on the total amount of plated lithium, but leads to apparent inhomogeneity in the distribution of plated lithium, with the differences between the concentrations of plated lithium in the upper and lower half of battery reach 57.61%, 103.06% and 159.68% of the average concentration in the whole battery during 1C charging at -5, 5 and 10°C, respectively. The temperature gradient also significantly influences the battery voltage relaxation profile related to the lithium stripping process. The characteristic plateau in the battery voltage relaxation profile disappears when the temperature gradient increases to larger than 6°C, leading to difficulty in detecting inhomogeneous lithium plating in large-format lithium batteries. The results presented in this study can provide helpful guidance for the design and management of large-format lithium-ion batteries. [Display omitted] [ABSTRACT FROM AUTHOR]

Summary: The inconsistency of the self‐discharge rate of each cell in series has an impact on the capacity of the battery pack, which is one of the best interpretations of the Cannikin Law. The cells with high self‐discharge rate in series determine the pack capacity. In addition, the cells with high self‐discharge rate usually means lower reliability and higher safety risk. Compared to traditional measurement methods, we previously proposed a method to calculate the self‐discharge rate by "pre‐parallel" equalization and then observing the current change of parallel branch. However, this method still has the problems of time‐consuming and difficult in batch sorting. This paper replaces the previous "pre‐parallel" operation with the initial voltage measurement of the cells. Combined with the established equivalent circuit model, the self‐discharge rate of each cell is calculated. The method further shortens the measurement time, which is suitable for sorting the cells in batches. The results show that the method can identify the absolute self‐discharge rate of each cell in a shorter time with relatively high accuracy. [ABSTRACT FROM AUTHOR]

Summary: Calendar life accounts for most of the lifetime of the power lithium‐ion battery applied in electric vehicles, and thus should be investigated in detail. In this paper, a mechanistic calendar aging model of lithium‐ion battery is developed by adding the solid electrolyte interface (SEI) growth side reaction to the pseudo‐two‐dimensional electrochemical model. The model can accurately simulate the capacity degradation and evolution of the charging voltage profiles of a Lix(NiCoMn)1/3O2 ‐ graphite battery during high‐temperature (55°C) storage under various states of charge. The model is further validated by comparing the electrode degradations inside the batteries after calendar aging. The model predicted electrode degradations are consistent with the results identified using a dual‐tank model and post‐mortem analysis, confirming that SEI growth is the dominant degradation mechanism. Based on the validated model, internal changes in electrodes' structure induced by SEI growth during calendaring aging are discussed. The anodes of the calendar‐aged batteries suffer from severe pore clogging and film resistance increase, resulting in apparent power performance degradation of lithium‐ion battery. Finally, modeling analysis is conducted to find possible solutions to mitigate the battery calendar aging process. Rational designs of SEI by reducing the porosity and solvent diffusivity inside the SEI and the kinetic rate constant turn out to be effective in improving the calendar lifetime of lithium‐ion batteries. Novelty Statement: Calendar life accounts for most of the lifetime of the lithium‐ion battery in electric vehicles. This study establishes a mechanistic calendar aging model of lithium‐ion battery considering solid electrolyte interface growth. The model can not only simulate the battery capacity degradation and the evolution of battery charging voltage profiles but also capture the internal degradation of the electrodes during the aging process. Modeling analysis is further conducted to find possible solutions to mitigate the calendar aging process of lithium‐ion battery. [ABSTRACT FROM AUTHOR]

Summary: Cell‐to‐pack (CTP) structure has been proposed for electric vehicles (EVs). However, massive heat will be generated under fast charging. To address the temperature control and thermal uniformity issues of CTP module under fast charging, experiments and computational fluid dynamics (CFD) analysis are carried out for a bottom liquid cooling plate based–CTP battery module. The impact of the channel height, channel width, coolant flow rate, and coolant temperature on the temperature and temperature difference are analyzed. A liquid cooling control method of exchanging the coolant inlet and outlet is proposed to optimize the temperature uniformity. Besides, the temperature uniformity enhancement by this control method at different intervals is compared. Results indicate that the flow rate and temperature positively affect the battery temperature; the maximum temperature can be reduced by 10.93% and 15.12%, respectively, under the same operations. However, the coolant temperature increment increases the maximum temperature difference by about 41.58%. Reversing flow enhances the cooling effect of conventional unidirectional flow of the CTP battery module under fast charging, especially for the thermal uniformity, which provides guidance for the battery thermal management system (BTMS) control under fast charging. [ABSTRACT FROM AUTHOR]

Electric vehicles are becoming increasingly prevalent as an effective solution to reduce resource scarcity and greenhouse gas emissions. As the core component of electric vehicles, lithium-ion batteries (LIBs) play a crucial role in energy storage and conversion. When LIBs are used in long-term service, it is essential to carefully consider the impact of modeling methods on both the environmental benefits and burdens associated with their usage. In this study, eight calculation models are chosen, and multiple environmental impacts of battery use-phase are compared based on life cycle assessment. The application of different computational models produced widely varying results, and previous conclusions regarding the environmental impacts of battery usage need to be revisited. Moreover, the different models during the battery use-phase exhibits an outstanding contribution to the life cycle carbon footprint and midpoint indicators. Sensitivity analysis reveals that battery efficiency is the most sensitive parameter to environmental indicators during use. Environmental impacts during the battery use-phase should consider the characteristics of different models to ensure the reliability and comparability of the study. • Midpoint/endpoint indicators and carbon footprint of eight calculation models during battery use-phase are investigated. • Carbon footprint at the national level depends on the source of electricity. • Electricity and models during the use-phase are critical factors influencing life cycle indicators. • Battery efficiency is the most sensitive parameter to environmental indicators during battery use-phase. [ABSTRACT FROM AUTHOR]

• The proposed method can detect lithium plating earlier in the charging process. • Compared with other methods, this method is more sensitive and simple. • Combined with the model, the source of lithium plating signal is analyzed. • The universality of the method is verified by model simulation. Lithium plating is one of the most severe problems threatening the durability and safety of lithium-ion batteries. Therefore, accurate detection of lithium plating is crucial for the health management and charging control of lithium-ion batteries. In this paper, an impedance-based method is proposed to detect lithium plating of lithium-ion battery by comparing the normalized charing internal resistance profiles. After verifying that the model-based method is effective and feasible, it was used to analyze the reason and internal mechanism of the detection signal compared with the more frequently used lithium plating detection procedures today. The proposed detection method is more sensitive and accurate and expected to detect lithium plating in real-time during the charging process. Finally, the model showed that the method remained effective and reliable for lithium plating detection when the charging conditions changed. This paper, through the analysis, simulation, and experimental verification of online lithium plating detection method based on charging internal resistance, provides innovative ideas for the subsequent development of lithium plating detection and stimulates potential technologies for future development. [ABSTRACT FROM AUTHOR]

The non-uniform performance of parallel cells in practical applications can lead to safety concerns, such as current overload, accelerated aging, and lithium plating. While overall voltage and current of parallel modules can be obtained, assessing the consistency among branch cells remains challenging due to the lack of direct branch current measurements. This paper introduces a method for accurately discerning the internal resistance consistency of parallel cells based on the relaxation voltage exhibited by the module. Analyzing the discharge process of parallel cells with different internal resistances, the link between internal resistance consistency and relaxation voltage recovery is discussed, and the effect of branch currents crossover is noted. A series of parallel modules are designed for experiments, and the results show that different internal resistance consistencies monotonically affect the recovery of the terminal voltage in the relaxation phase when discharge is stopped before the parallel branch currents crossover. Thus, the internal resistance consistency can be judged by the behavior of the relaxation voltage. A faster terminal voltage recovery of the module in the relaxation phase indicates better internal resistance consistency. This method requires only terminal voltage data and no additional sensors, which provides superior economy and practicality for real applications. • Proposed a current sensorless method for identifying the internal resistance consistency of parallel cells. • The relaxation voltage of parallel module reflects the internal resistance consistency of parallel cells within the module. • Validated the method through simulations and experiments. • Suggested a practical application workflow for the method. [ABSTRACT FROM AUTHOR]

The overall performance of lithium-ion battery is determined by the innovation of material and structure of the battery, while it is significantly dependent on the progress of the electrode manufacturing process and relevant equipment and technology. Battery manufacturers have been generally employing the exhaustive method for the trials of the electrode process development that pertains to different systems, and the process simulation technology still has a large space for development. To comply with the development trend of high-quality battery manufacturing and digital intelligent upgrading industry, the existing research status of process simulation for electrode manufacturing is systematically summarized in this paper from the perspectives of macro battery manufacturing equipment and micro battery electrode structure. Moreover, the mechanism research, structural development, and application prospect of the respective process simulation technology are analyzed. The limitations of existing research and future development trends are further highlighted to provide theoretical reference for optimizing the manufacturing process and enhancing the comprehensive performance of lithium-ion batteries. [Display omitted] • Systematic review of lithium-ion battery electrode process simulation. • Analyzing simulation techniques, revealing mechanisms, evolution, applications. • Critically reviewing research limits, providing insightful projections. [ABSTRACT FROM AUTHOR]

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. [ABSTRACT FROM AUTHOR]

Power capability of lithium-ion batteries is strongly correlated with electric vehicle's accelerating and braking performance. However, the estimate of state-of-power relies highly on battery models, whose accuracy usually increases with the complexity. We here propose a simple and accurate frequency-dependent integer-order model for battery state-of-power estimation. First, a random search-pseudo gradient descent algorithm is proposed to identify the parameters of our model from electrochemical impedance spectroscopy in the frequency domain. Then, the proposed model is mathematically derived in the time domain. Next, two strategies are developed to estimate battery state-of-power under different constraints — using particle swarm optimization and direct inversion algorithms. Finally, our method is experimentally verified: the proposed frequency-dependent model shares similar complexity compared with the conventional integer-order model, while its accuracy is competitive to that of the fractional-order model. With such a simple and accurate model, our state-of-power estimation error is 90% smaller than that based on the conventional integer order model, and the computational time is 99.8% lower than that corresponds to the fractional-order model. Since the proposed method is developed upon the conventional integer-order model, it has a strong potential for real-life application and can be easily integrated into the existing battery management systems. • Replace the fractional-order model with the Frequency-dependent integer- order model. • Models are derived and analyzed in both time and frequency domains. • The frequency-dependent model supports direct state-of-power calculations. • High accuracy and low computational burden. • Algorithms are verified with experimental-based state-of-power ground truth. [ABSTRACT FROM AUTHOR]

Aging modes analysis of lithium-ion batteries plays a crucial role in battery health management. The present studies for battery aging modes analysis are mainly based on mechanistic models or electrochemical models. However, most of the parameters of these models need to be measured offline, which adds difficulties to actual vehicle applications. Therefore, this paper proposes a method for analyzing and predicting battery aging modes based on a transfer learning method. Aging modes data of experimental batteries and electric vehicle batteries (EVBs) are obtained by an enhanced dual-tank model. Then, a transfer learning model based on long and short-term memory neural network is trained to achieve the analysis of EVBs aging modes by using the experimental battery aging data as the source data and the aging data of EVBs as the target data. The results show that the estimation error for the EVBs aging parameters is less than 6% and the predicted error for the next 5 months is less than 4%. The method offers a new idea for EVBs online aging modes analysis and prediction. • A EVBs voltage reconstruction method based on the internal resistance compensation method is proposed. • An enhanced dual-tank model can be used to estimate battery capacity while analyzing battery aging modes. • Transfer learning based model is used for aging parameters estimation of EVBs. • An unsupervised learning based transfer model is used for aging parameter prediction in EVBs. [ABSTRACT FROM AUTHOR]