For the safe and reliable operation of battery systems, accurate and fast fault diagnosis is vitally important to prevent accidents. However, some faults exhibit similar physical characteristics and remain undetected due to their minor scale, posing significant challenges for fault diagnosis. In this work, an efficient diagnostic method is proposed to detect and isolate external short circuit and connection faults. A specialized one-to-one voltage measurement topology is employed for fault isolation. Voltage difference variations within a time window are utilized to preprocess sampling voltages and extract fault features. A diagnostic model is then established based on the voltage difference variations and independent component analysis. Squared prediction error (SPE) and Hotelling's T2 statistics are combined to complement each other in fault detection. To address the issue of misdiagnosis in classical contribution plots, an improved contribution to SPE is proposed based on relative contributions, and the corresponding contribution plot is selected for fault isolation based on the fault detection results. Finally, the proposed diagnostic method is validated on a series-connected battery pack under various fault scenarios. The results demonstrate that the proposed method achieves reliable and fast fault diagnosis.

Battery energy storage systems bolster power grids” absorption capacity, however, battery safety issues remain a formidable challenge. Timely and precise fault diagnosis, coupled with early-stage fault warnings., is crucial. This study introduces an eigen decomposition-based multi-fault diagnosis approach for lithium-ion battery packs, enabling online diagnosis of short circuits, electrical connection faults, and voltage sensor malfunctions. By incorporating an interleaved measurement topology, precise fault type differentiation is achieved. Eigenvector matching analysis is employed to increase sensitivity to fault characteristics and enhance robustness. The interleaved topology can be seamlessly integrated using common voltage measurement solutions, eliminating the need for additional design complexities, while sensor number redundancy enhances fault tolerance of battery management systems (BMS). A cloud-side collaboration method is proposed, where the BMS functions as an edge device for specific data computations, while the parameters are fine-tuned by the server through big data analytics. This approach circumvents cumber-some server calculations, thereby curbing server cost escalation. The edge computing process is divided into two steps, with partial calculations often sufficient to evaluate battery safety, thus reducing the computational load on edge devices. Several battery tests are conducted, and the results confirm the method's capability, feasibility, and validity in early-stage fault diagnosis.

O3-type NaNi1/3Fe1/3Mn1/3O2 as a promising cathode material for sodium-ion batteries suffers from poor cycle life and air stability. Elemental doping and surface coating are effective strategies to improve the stability of underlying cathode materials. In this work, O3-type NaNi1/3Fe1/3Mn1/3O2 is modified with Zr via two distinct synthesis routes: introducing Zr during the precursor stage or after the cathode calcination. Calcination of a Zr (NO3)4-treated Ni1/3Fe1/3Mn1/3(OH)2 precursor forms Zr-doped material, whereas processing of Zr(NO3)4-treated NaNi1/3Fe1/3Mn1/3O2 forms a Na2ZrO3-coated and trace Zr4+-doped material. Zr4+ doping enhances the structure stability and Na+ diffusion, resulting in the improved cycling stability and significantly enhanced rate capability of 81 mAh & sdot;g-1 at 10C compared to that of 44 mAh & sdot;g-1 for pristine NaNi1/3Fe1/3Mn1/3O2. Meanwhile, the Na2ZrO3-coated NaNi1/3Fe1/3Mn1/3O2 delivers highest cycling stability of 69% and 78% retention after 100 cycles at 1C and 3C, respectively, by suppressing interfacial side reactions. This work demonstrates that the synthesis route dictates the dominant mechanism, offering a rational design principle for optimizing layered oxide cathodes.

In the design of traditional energy management strategies for energy storage system clusters in response to grid power demand, the influence of cascade converter on systematic energy consumption characteristics are generally ignored, making it unable to achieve energy loss optimization within service life. In this paper, a high-order accurate energy consumption characteristic model is established by comprehensively considering the power efficiency characteristics of cascade converters, and a real-time analytics based optimal energy management strategy is proposed. It is inspired by Taylor expansion based order-reduced iterative solving, which can analytically characterize the correlation between systematic energy loss and power distribution among energy storage systems. Based on the hardware-in-the-loop simulation, the results demonstrate that the accuracy of high-order energy consumption characteristic modeling for energy storage systems is up to 99.8%, and the real-time analytics based systematic energy loss optimization can be ensured. Furthermore, around 28.2% improvement in the steady-state state-of-charge balancing performance and at least 4% service life enhancement can be guaranteed in comparison with the typical state-consensus based energy management strategy. In the meanwhile, the online fault-tolerance functionality is inherent as it is indeed the special case of the normal operation.

Because of their advantages of high energy and power density, low self-discharge rate, and long lifespan, lithium-ion batteries (LIBs) have been widely used in many applications such as electric vehicles, energy storage systems, smart grids, etc. However, lithium-ion battery systems (LIBSs) frequently malfunction because of complex working conditions, harsh operating environment, battery inconsistency, and inherent defects in battery cells. Thus, safety of LIBSs has become a prominent problem and has attracted wide attention. Therefore, efficient and accurate fault diagnosis for LIBs is very important. This paper provides a comprehensive review of the latest research progress in fault diagnosis for LIBs. First, the types of battery faults are comprehensively introduced and the characteristics of each fault are analyzed. Then, the fault diagnosis methods are systematically elaborated, including model-based, data processing-based, machine learning-based and knowledge-based methods. The latest research is discussed and existing issues and challenges are presented, while future developments are also prospected. The aim is to promote further researches into efficient and advanced fault diagnosis methods for more reliable and safer LIBs.

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.

In order to avoid thermal runaway as well as ensure safe and stable operation of battery system, it is vital to perform rapid and accurate diagnosis for early faults with similar characteristics. In this paper, a short circuit and connection fault diagnosis method that is applicable to traditional one-to-one voltage sensor topology is proposed based on differential voltage changes (DVCs) and independent component analysis. Voltage measurements is pretreated to obtain DVCs for extracting fault signature and reducing the detrimental effects of inconsistency, interference, etc. to a certain extent. Independent component analysis is combined with the DVC vector to establish diagnostic model and diagnose faults online. Fault detection and isolation is achieved using squared prediction error, and considering that traditional contributions cannot accurately determine fault location, an improved contribution calculation is introduced to localize the faulty cell. Experimental verifications show that the proposed method can timely detect and localize faults, meanwhile it can distinguish short circuit with connection fault.

In order to solve the problems of reactive power optimization in distribution networks which includes the high proportion of distributed new energy power generation, this paper proposes a multi time scale reactive power optimization method. Firstly, on the day-ahead scale, the slow motion control variables were optimized with the objective function of minimizing the expected network loss, and a master-slave classification period decoupling strategy was proposed to transform the dynamic optimization of discrete variables into Static optimization to simplify the optimization process; Secondly, on the 15 minute scale, rolling optimization of reactive power output of new energy is adopted, with the goal of minimizing the system voltage deviations; Finally, in real time, reactive power output of new energy is optimized at any time with the goal of maintaining a constant voltage at the grid connection point within each time window to suppress voltage stability issues caused by various random fluctuations. Finally, the effectiveness and feasibility of the proposed method were verified using the modified IEEE33 node as an example.

The basic principle of sensorless technology for interior permanent magnet synchronous motors (IPMSM) suitable for zero and low speed operation is to detect the convex pole of the motor, and since the convex modulation signal of the motor contains position information, the position information can be estimated by different signal detection and separation methods. Usually, the pulsating high-frequency (HF) voltage signal injection method is used to extract the position signal by filters, but since the filter affects the dynamic performance of the system, a rotor position observation method based on the second-order generalized integrator (SOGI) is proposed. Through simulation and experimental comparison, the results show that the method improves the observation effect of the system.

Fault diagnosis for battery circuit is particularly important for the safe management of electric vehicles. Previous correlation based fault diagnosis method only detect some faults, ignores the coupled faults, load connection faults and the problem of current data submerged. In this paper, a multi-fault online diagnosis approach combining a non-redundant measurement topology and weighted Pearson correlation coefficient (WPCC) is adopted to detect various circuit faults by weighted measured data with different forgetting factors. The main advantages are: 1) With adding the connected resistances between the battery pack and the load, the non-redundant measurement topology contains a current sensor and the same number of voltage sensors as those of the battery cells without adding complexity to the system. 2) By adding different weights with bigger forgetting factor to more recent data, a period signal aided WPCC approach is adopted to forget historical data and stress the recent data, so as to online detect the circuit faults. 3) Different from the previous same kind of fault judgement idea, the comprehensive judgement rule are used to online judge the battery abuse faults, connection faults, sensor faults, adjacent homogeneous faults and adjacent hybrid faults. The experiment results show that the investigated method can distinguish and locate the above faults accurately.

Lithium-ion batteries have become the fastest-growing energy storage equipment available for extrinsic and intrinsic reasons. State of Charge (SOC) is one of the lithium-ion batteries' most critical performance indicators, reflecting the remaining capacity. An accurate and stable estimate of SOC is critical for any lithium-ion battery. This paper proposes a hybrid method to achieve stable and real-time battery SOC estimation at different temperatures, composed of an Improved Bidirectional Gated Recurrent Unit (IBGRU) network and Unscented Kalman filtering (UKF). The proposed method is experimentally validated using data from UDDS and US06 driving cycles. The verification results show that the method can adapt to various working conditions and obtain good estimation accuracy and robustness, with MAE and RMSE less than 0.83% and 1.12%, respectively. After transfer learning, the method can also be applied to new lithium-ion batteries and achieve good estimation performance at new temperature conditions. The maximum errors are 4.98% and 5.76% at 25 degrees C and -10 degrees C, respectively. Therefore, the IBGRU-UKF method can achieve a more accurate and stable SOC estimation with good expansion performance for different lithium-ion batteries.

The state of health (SOH) is critical to the efficient and reliable use of lithium-ion batteries (LIBs). Recently, the SOH estimation method based on electrochemical impedance spectroscopy (EIS) has been proven effective. In response to different practical applications, two models for SOH estimation are proposed in this paper. Aiming at based on the equivalent circuit model (ECM) method, a variety of ECMs are proposed. Used EIS to predict the ECM, the results show that the improved method ensures the correctness of the ECM and improves the estimation results of SOH. Aiming at a data-driven algorithm, proposes a convolution neural network (CNN) to process EIS data which can not only extract the key points but also simplifies the complexity of manual feature extraction. The bidirectional long short-term memory (BiLSTM) model was used for serial regression prediction. Moreover, the improved Particle Swarm Optimization (IPSO) algorithm is proposed to optimize the model. Comparing the improved model (IPSO-CNN-BiLSTM) with the traditional PSO-CNN-BiLSTM, CNN-BiLSTM and LSTM models, the prediction results are improved by 13.6%, 93.75% and 94.8%, respectively. Besides that, the two proposed methods are 27% and 35% better than the existing gaussion process regression (GPR) model, which indicates that the proposed improved methods are more flexible for SOH estimation with higher precision.

The bullet trains use electric traction. As a key component, the electrical system provides power for the operation of the whole vehicle. The auxiliary power unit (APU) provides power for the electric multiple unit (EMU) electrical system. Nickel-cadmium batteries are used as backup power supply. With the train running, the storage battery inevitably appears a certain degree of aging. In order to improve the service life and safety of EMU batteries, realize the omnidirectional monitoring and management of batteries, and establish the life prediction model and optimal replacement strategy of batteries, this paper designs a battery management system (BMS). The BMS adopts a master-slave structure. The master module is based on S32K144 automotive microcontroller, and the slave module is based on MC33771 in AFE chips. The BMS can collect and analyze the operation parameters of the battery in real time. The ampere-hour integration method and Extended Kalman Filter (EKF) were used to estimate the SOC (state of charge) of the battery in real time. Simulation in Simulink software and experimental verification show that the estimation error can be within 2%.

Remaining useful life shows extraordinary function in guiding the timely replacement of supercapacitors that reach the service life limit, which has great significance to the security and stability of the energy storage system. In order to more accurately predict the remaining useful life of supercapacitors so as to ensure the reliability of the whole supercapacitor bank, a temporal convolutional network is used. Among them, a residual block can solve the problems of gradient explosion and gradient disappearance, which are widespread in the recurrent neural network. Early stopping technology is used to avoid overfitting, and the Adam algorithm was used to optimize the process of parameter adjustment of the temporal convolutional network. The stability and accuracy of the model prediction were verified by using the capacity attenuation dataset of supercapacitors under different experimental conditions. Meanwhile, to verify the generalization ability of the model, the datasets of supercapacitors at different working conditions without training are input into the temporal convolutional network model. Simulation shows that the temporal convolutional network model exhibits strong robustness and high accuracy in predicting the remaining useful life of supercapacitors.

Lithium-ion batteries have been used in all aspects of life for environmental and resource reasons. The accurate estimation of the state of charge (SOC) ensures the proper operation of the battery. However, few methods have focused on the problem of SOC estimation at low temperatures. A hybrid method based on the CNN-BWGRU network is proposed in this paper. The method optimizes the influence of battery information on the results through a “multi-moment input” structure and the bidirectional network. The convolutional neural network (CNN) is used to learn the feature parameters in the input. The bidirectional weighted gated recurrent unit (BWGRU) can improve the fitting performance of the network at low temperatures by changing the weights. The proposed network has a strong generalization capability, estimation accuracy, and robustness. SOC estimation is performed under different conditions to verify the plausibility of the network. The experimental results show that the method has higher accuracy and stability than other networks. In addition, the proposed method can overcome the effect of different initial SOC on the estimation results. Therefore, the CNN-BWGRU network provides a new method in battery SOC estimation while providing helpful guidance for safe and stable battery operation in natural environments.

针对传统粒子滤波估算电池荷电状态(SOC)出现的粒子贫化问题,本工作提出了一种改进蝙蝠算法(IBA)优化粒子滤波(PF)来估算SOC的方法.将粒子表征为蝙蝠个体,模仿蝙蝠族群捕食过程,解决粒子滤波技术中粒子贫化问题;结合二阶Thevenin电池模型构建电池状态空间理论模型并对电池进行相关参数辨识;利用IBA-PF算法与标准PF算法在脉冲电流工况和DST工况电流下进行SOC估算.实验结果证明,与传统的PF算法比较,基于IBA-PF的锂电池SOC估算精度在2％以内,对非线性和非高斯特性的锂电池SOC估算具有良好的适应性和稳定性.

针对锂电池老化机理复杂、电池健康状态(SOH)估算不准确问题,提出了一种新型电池SOH估计方法.首先,建立基于改进蚁群算法优化支持向量回归(IACA-SVR)的电池SOH预测模型,以电池放电过程平均电压和平均温度为输入变量,模拟电池老化.其次,利用样本数据训练SVR,并用IACA来优化SVR的参数.在数据集上的实验结果表明,经过优化的SVR预测结果与遗传算法SVR(GA-SVR)预测结果相比更为精确,稳定性更好,从而验证了IACA-SVR预测模型的可行性,能为电动汽车锂电池的安全使用提供准确电池数据.

为了准确估计电池当前健康状态（SOH）,提出一种基于改进粒子滤波算法的电池SOH在线评估方法。首先针对传统萤火虫优化算法的不足,提出了一种改进萤火虫算法替代传统粒子滤波的重采样,然后从锂离子电池工作时的可测参数中提取在线健康指标（HI）,建立HI与SOH之间的映射模型,并将其应用于状态空间模型的观测。利用马里兰大学先进寿命周期工程中心（CALCE）公布的试验测量数据进行验证,结果表明,该方法对具有非线性和非高斯特性的锂离子电池降解过程的状态估计具有良好的适应性。

为提高锂离子电池荷电状态(state of charge,SOC)值和健康状态(state of health,SOH)值的估算精度,基于二阶戴维南等效电池模型,提出双自适应无迹卡尔曼滤波(double adaptive unscented Kalman filter,DAUKF)算法.通过AUKF1和AUKF2这2个滤波器,可以同时计算出电池的SOC值和电池内阻,内阻既可以更新电池的模型参数,又可依靠函数关系,估算出电池的SOH值.仿真结果表明,DAUKF能够准确估算出SOC值和SOH值,精度保持在2％以内,由此验证了该方法的可行性和精确性.