Estimation of power battery SOC based on firefly BP neural network

针对BP神经网络算法对电动汽车电池荷电状态（state of charge,SOC）估算的缺陷,提出一种基于萤火虫（firefly algorithm,FA）神经网络的SOC估算方法。以磷酸铁锂电池为测试对象,在ARBIN公司生产的EVTS电动车动力电池测试系统装置上进行测试,收集锂电池的各项性能参数。采用端电压和放电电流作为输入参数,SOC作为输出参数,建立FA-BP神经网络模型,用于估算锂离子电池充放电过程中的任一状态下的SOC。仿真实验结果表明,与现有的BP神经网络估算方法相比,基于FA-BP神经网络的锂电池SOC估算方法准确度高,具备很好的实用性。     

Equalization of series connected lithium-ion batteries based on back propagation neural network and fuzzy logic control

In this article, a nondissipative equalization scheme is proposed to reduce the inconsistency of series connected lithium-ion batteries. An improved Buck-Boost equalization circuit is designed, in which the series connected batteries can form a circular energy loop, equalization speed is improved, and modularization is facilitated. This article use voltage and state of charge (SOC) together as equalization variables according to the characteristics of open-circuit voltage (OCV)-SOC curve of lithium-ion battery. The second-order RC equivalent circuit model and back propagation neural network are used to estimate the SOC of lithium-ion battery. Fuzzy logic control (FLC) is used to adjust the equalization current dynamically to reduce equalization time and improve efficiency. Simulation results show that the traditional Buck-Boost equalization circuit and the improved Buck-Boost equalization circuit are compared, and the equalization time of the latter is reduced by 34%. Compared with mean-difference algorithm, the equalization time of FLC is decreased by 49% and the energy efficiency is improved by 4.88% under static, charging and discharging conditions. In addition, the proposed equalization scheme reduces the maximum SOC deviation to 0.39%, effectively reducing the inconsistency of batteries.     

Improved state-of-the-art look-up table method for charge state estimation of PSO-RBF model

为了解决纯电动汽车在SOC预测时易受电流波动、工况非线性等因素影响,提出了一种针对锂电池SOC进行动态预测的方法。首先,对粒子群聚类算法的参数组合进行优选并结合优选结果对径向基函数（RBF）神经网络进行改进。然后,通过分析电池在不同工作状态下的特性,将电池分为充电、静置、放电三个状态。针对电池所处的工作状态采取不同的策略对SOC进行预测。在电池放电阶段使用经过改进的PSO-RBF算法对SOC进行动态预测。在电池静置及充电状态使用二分查表法,将考虑温度漂移的开路电压曲线及充电时电流节点突变曲线制成二维数组表,利用制作的二维数组表对SOC的值进行修正。从而减小系统响应时间,同时预测提升精度。实验结果表明,该预测修正模型最大误差约为1.9%,验证了方法的有效性。     

Online management of lithium-ion battery based on time-triggered controller area network for fuel-cell hybrid vehicle applications

This paper introduces a state of charge (SOC) estimation algorithm that was implemented for an automotive lithium-ion battery system used in fuel-cell hybrid vehicles (FCHVs). The proposed online control strategy for the lithium-ion battery, based on the Ah current integration method and time-triggered controller area network (TTCAN), incorporates a signal filter and adaptive modifying concepts to estimate the Li₂MnO₄ battery SOC in a timely manner. To verify the effectiveness of the proposed control algorithm, road test experimentation was conducted with an FCHV using the proposed SOC estimation algorithm. It was confirmed that the control technique can be used to effectively manage the lithium-ion battery and conveniently estimate the SOC.     

SOC estimation method of power battery based on BP artificial neural network

电池剩余电量（SOC）的估算是电池管理系统中的关键技术之一,在众多估算方法中,神经网络在估算的准确性及鲁棒性上具有明显优势。庞大的数据量是获得SOC精确值的重要因素。针对以上问题,研究提出了基于BP人工神经网络的动力电池SOC估算方法,以某型号整包电池作为实验对象,通过对电池电压、电流、内阻及温度的数据采集,获得海量数据。建立电池的等效电路模型,考虑电池极化、充放电倍率及温度的影响对初始数据进行修正。基于MATLAB平台建立BP人工神经网络模型,数据修正后用于网络模型的训练,并验证了模型的可行性。将模型用于实验数据的预测,通过函数拟合实现了SOC的估算。最后,通过对比SOC的预测值与实际测量值,最终证明建立的人工神经网络模型对SOC估算的有效性。     

An enhanced temperature-dependent model and state-of-charge estimation for a Li-Ion battery using extended Kalman filter

Development of high-fidelity mathematical models and state-of-charge (SOC) estimation of Li-ion battery becomes a significant challenge when the temperature effects are considered. In this paper, we propose an enhanced temperature-dependent equivalent circuit model for a Li-ion battery and applied it for battery parameters estimation and model validation, as well as SOC estimation. First, the new battery model is elaborated, including a newly integrated resistance-capacitor structure, a static hysteresis voltage and a temperature compensation voltage term. The forgetting factor least square approach is utilized to realize the parameter identification. Next, the proposed battery model is employed to estimate battery SOC by incorporating the extended Kalman filter algorithm. Finally, simulation results are provided to demonstrate the superior performance of the proposed battery model in comparison with the common first-order Thevenin temperature model. Compared with Thevenin model, the maximal values of relative reconstruction error and root mean squared error with the proposed battery model are decreased by about 33.3% and 50.0%, respectively, for the battery terminal output voltage, 50.0% and 53.0%, respectively, for the SOC estimation, under three different test profiles.     

Neural network-based learning and estimation of battery state-of-charge: A comparison study between direct and indirect methodology

Faced with the ever-increasing urban environmental pollution, the electric vehicles (EVs) have received increasing attention in the automotive industry. Lithium-ion batteries, serving as electrochemical power storage, have been extensively used in EVs because of the lightweight, no local pollution and high power density. The increasing awareness on the safe operation and reliability of the battery requires an efficient battery management system (BMS), among the parameters monitored by which, state-of-charge (SOC) is critical in preventing overcharge, deep discharge, and irreversible damage. This article investigates the neural network (NN)-based modeling, learning, and estimation of SOC by comparing two different methodologies, that is, direct structure with SOC as network output and indirect structure with voltage as output. Firstly, the nonlinear autoregressive exogenous neural network (NARX-NN) is introduced, in which SOC is directly deemed as an NN output for learning and estimation. Secondly, a radial basis function (RBF)-based NN with unscented Kalman filter (RBFNN-UKF) is proposed, in which the terminal voltage is used as output. Instead, SOC is deemed as an internal state which would be estimated indirectly based on the feedback error of voltage. Experimental results demonstrate that both estimators can achieve accurate SOC estimation for regular cases, in spite of the inaccurate initial conditions. However, the direct NN structure is revealed as not capable of dealing with the cases with sensor bias, which, however, can be well accommodated in the indirect structure by extending the sensor bias as an augmented state. Benefiting from the uncertainty augmentation and feedback compensation, the indirect RBFNN-UKF shows superiority over the direct estimation in the practical experiments, depicting a promising prospect in the future onboard EV-BMS application.     

State of charge estimation method based on the extended Kalman filter algorithm with consideration of time-varying battery parameters

In developing battery management systems, estimating state-of-charge (SOC) is important yet challenging. Compared with traditional SOC estimation methods (eg, the ampere-hour integration method), extended Kalman filter (EKF) algorithm does not depend on the initial value of SOC and has no accumulated error, which is suitable for the actual working condition of electric vehicles. EKF is a model-based algorithm; the accuracy of SOC estimated by this algorithm was greatly influenced by the accuracy of battery model and model parameters. The parameters of battery change with many factors and exhibit strong nonlinearity and time variance. Typical EKF algorithm approximates battery as a linear, time-invariant system; however, this approach introduces estimation errors. To minimize such errors, previous studies have focused on improving the accuracy of identifying battery parameters. Although studies on battery model with time-varying parameters have been carried out, few have studied the combination of time-varying battery parameters and EKF algorithm. A SOC estimation method that combines time-varying battery parameters with EKF algorithm is proposed to improve the accuracy of SOC estimation. Battery parameter data were obtained experimentally under different temperatures, SOC levels, and discharge rates. The results of parameter identification are made into a data table, and the battery parameters in the EKF system matrix are updated by looking up the data in the table. Simulation and experimental results shown that, average error of SOC estimated by the proposed algorithm is 2.39% under 0.9 C constant current discharge and 2.4% under 1.3 C, which is 1.91% and 2.35% lower than that of EKF algorithm with fixed battery parameters. Under intermittent discharge with constant current (1.1 C) and capacity (10%), the average error of SOC estimated by the proposed algorithm is 1.4%, which is 0.3% lower than that of EKF algorithm with fixed battery parameters. The average error of SOC estimated by the proposed algorithm under the New European Driving Cycle (NEDC) is 1.6%, which is 0.2% lower than that of EKF algorithm with fixed battery parameters. Relative to the EKF algorithm with fixed battery parameters, the proposed EFK algorithm with time-varying battery parameters yields higher accuracy.     

A wavelet transform‐adaptive unscented Kalman filter approach for state of charge estimation of LiFePo4 battery

LiFePo₄ battery is widely used in electric vehicles; however, its flatness and hysteresis of the open‐circuit voltage curve pose a big challenge to precise state of charge (SOC) estimation. The issue is discussed and addressed in this paper. First, a cell model with hysteresis is built to describe real‐time dynamic characteristics of the LiFePo₄ battery. Second, the model parameters and SOC are estimated independently to avoid the possibility of cross interference between them. For model identification, an adaptive unscented Kalman filter (AUKF) algorithm is used to identify the cell parameters as they change slowly. While SOC could change rapidly, wavelet transform AUKF algorithm is put forward to estimate SOC. In the novel algorithm, the measurement noise can be estimated and updated online. Finally, the performance of the proposed method is verified under dynamic current condition. The experimental results show that estimated value based on the proposed method is more accurate than unscented Kalman filter‐based method and AUKF‐based algorithm. Meanwhile, the proposed estimator also has the merits of fast convergence and good robustness against the initialization uncertainty.     

A power allocation strategy of a hybrid energy storage system for a PV grid-connected system

针对光伏并网系统中光伏微电源出力的波动性和间歇性,将蓄电池和超级电容器构成的混合储能系统HESS（hybrid energy storage system）应用到光伏并网系统中可以实现光伏功率平滑、能量平衡以及提高并网电能质量。在同时考虑蓄电池的功率上限和超级电容的荷电状态（SOC）的情况下,对混合储能系统提出了基于超级电容SOC的功率分配策略;该策略以超级电容的SOC和功率分配单元的输出功率作为参考值,对混合储能系统充放电过程进行设计。超级电容和蓄电池以Bi-direction DC/DC变换器与500 V直流母线连接,其中超级电容通过双闭环控制策略对直流母线电压进行控制。仿真结果表明,所提功率分配策略能对混合储能系统功率合理分配,而且实现了单位功率因数并网,稳定了直流母线电压。     

基于PNGV模型储能锂电池参数辨识及SOC估算研究

锂电池因具有比能量高、循环寿命长、对环境无污染等优点,在储能系统中已逐渐得到应用.准确估算锂电池的荷电状态(SOC)可防止电池过充、过放,保障电池安全、充分地使用.为了精确估算储能锂电池SOC,基于PNGV(partnership for a new generation of vehicles)电池等效模型,利用递推最小二乘法(RLS)对模型参数进行在线辨识和实时修正,增强了系统的适应性.结合安时法、开路电压法和PNGV模型,提出了一种实时在线修正SOC算法.根据实验数据,建立了仿真模型,以验算模型和SOC估算算法的精度.仿真结果表明,PNGV模型能真实地模拟电池特性,且能有效地提高SOC估算精度,适合长时间在线估算储能锂电池的SOC.     

A new state‐of‐charge estimation method for electric vehicle lithium‐ion batteries based on multiple input parameter fitting model

The estimation of state‐of‐charge (SOC) is crucial to determine the remaining capacity of the Lithium‐Ion battery, and thus plays an important role in many electric vehicle control and energy storage management problems. The accuracy of the estimated SOC depends mostly on the accuracy of the battery model, which is mainly affected by factors like temperature, State of Health (SOH), and chemical reactions. Also many characteristic parameters of the battery cell, such as the output voltage, the internal resistance and so on, have close relations with SOC. Battery models are often identified by a large amount of experiments under different SOCs and temperatures. To resolve this difficulty and also improve modeling accuracy, a multiple input parameter fitting model of the Lithium‐Ion battery and the factors that would affect the accuracy of the battery model are derived from the Nernst equation in this paper. Statistics theory is applied to obtain a more accurate battery model while using less measurement data. The relevant parameters can be calculated by data fitting through measurement on factors like continuously changing temperatures. From the obtained battery model, Extended Kalman Filter algorithm is applied to estimate the SOC. Finally, simulation and experimental results are given to illustrate the advantage of the proposed SOC estimation method. It is found that the proposed SOC estimation method always satisfies the precision requirement in the relevant Standards under different environmental temperatures. Particularly, the SOC estimation accuracy can be improved by 14% under low temperatures below 0 °C compared with existing methods. Copyright © 2017 John Wiley & Sons, Ltd.     

Research on parameter identification and state of charge estimation of improved equivalent circuit model of Li-ion battery based on temperature effects for battery thermal management

The performance and parameters of Li-ion battery are greatly affected by temperature. As a significant battery parameter, state of charge (SOC) is affected by temperature during the estimation process. In this paper, an improved equivalent circuit model (IECM) considering the influence of ambient temperatures and battery surface temperature (BST) on battery parameters based on second-order RC model have been proposed. The exponential function fitting (EFF) method was used to identify battery model parameters at 5 ambient temperatures including −10°C, 0°C, 10°C, 25°C and 40°C, fitting the relationship between internal resistance and BST. Then, the SOC of the IECM was estimated based on the extended Kalman filter (EKF) algorithm. Using the result calculated by the Ampere-hour integration method as the standard, the data of battery under open circuit voltage (OCV) test profile and dynamic stress test (DST) profile at different ambient temperatures has been compared with the ordinary second-order RC model, and the advantages of the SOC estimation accuracy with IECM was verified. The numerical results showed that the IECM can improve the estimation accuracy of battery SOC under different operating conditions.     

Model-based power control strategy development of a fuel cell hybrid vehicle

An integrated procedure for math modeling and power control strategy design for a fuel cell hybrid vehicle (FCHV) is presented in this paper. Dynamic math model of the powertrain is constructed firstly, which includes four modules: fuel cell engine, DC/DC inverter, motor-driver, and power battery. Based on the mathematic model, a power control principle is designed, which uses full-states closed-loop feedback algorithm. To implement full-states feedback, a Luenberger state observer is designed to estimate open circuit voltage (OCV) of the battery, which make the control principle not sensitive to the battery SOC (state of charge) estimated error. Full-states feedback controller is then designed through analyzing step responding of the powertrain and test data. At last of the paper, the results of simulation and field test are illustrated. The results show that the power control strategy designed takes into account the performance and economy characteristics of components of the FCHV powertrain and achieves the control object excellently.     

Discrimination of Li-ion batteries based on Hamming network using discharging-charging voltage pattern recognition for improved state-of-charge estimation

Differences in electrochemical characteristics among Li-ion batteries and factors such as temperature and ageing result in erroneous state-of-charge (SoC) estimation when using the existing extended Kalman filter (EKF) algorithm. This study presents an application of the Hamming neural network to the identification of suitable battery model parameters for improved SoC estimation. The discharging-charging voltage (DCV) patterns of ten fresh Li-ion batteries are measured, together with the battery parameters, as representative patterns. Through statistical analysis, the Hamming network is applied for identification of the representative DCV pattern that matches most closely of the pattern of the arbitrary battery to be measured. Model parameters of the representative battery are then applied to estimate the SoC of the arbitrary battery using the EKF. This avoids the need for repeated parameter measurement. Using model parameters selected by the proposed method, all SoC estimates (off-line and on-line) based on the EKF are within ±5% of the values estimated by ampere-hour counting.     

Experimental validation for Li-ion battery modeling using Extended Kalman Filters

The battery management systems (BMS) is an essential emerging component of both electric and hybrid electric vehicles (HEV) alongside with modern power systems. With the BMS integration, safe and reliable battery operation can be guaranteed through the accurate determination of the battery state of charge (SOC), its state of health (SOH) and the instantaneous available power. Therefore, undesired power fade and capacity loss problems can be avoided. Because of the electrochemical actions inside the battery, such emerging storage energy technology acts differently with operating and environment condition variations. Consequently, the SOC estimation mechanism should cope with the probable changes and uncertainties in the battery characteristics to ensure a permanent precise SOC determination over the battery lifetime.This paper aims to study and design the BMS for the Li-ion batteries. For this purpose, the system mathematical equations are presented. Then, the battery electrical model is developed. By imposing known charge/discharge current signals, all the parameters of such electrical model are identified using voltage drop measurements. Then, the extended kalman filter (EKF) methodology is employed to this nonlinear system to determine the most convenient battery SOC. This methodology is experimentally implemented using C language through micro-controller. The proposed BMS technique based on EKF is experimentally validated to determine the battery SOC values correlated to those reached by the Coulomb counting method with acceptable small errors.     

Inflection point Ah-total integration method for real-time integration to correct lithium battery SOC

对18650及26650磷酸铁锂电池的充放电电流、电压等数据分析表明：在电池循环老化过程中，虽然容量电压曲线两端曲率最大（拐点）处的SOC值有所变化，但是其电压保持不变。因此在估算SOC过程中，当放电电压达到拐点电压时，将此时的SOC修正为对应的拐点SOC，可以一定程度上优化安时积分法由于初始SOC而估算不准的问题。在此基础上提出一种新型的拐点修正安时积分算法，综合考虑温度、充放倍率、循环老化等因素对SOC估算精度的影响，引入充放电曲线拐点概念，建立SOC实时估算数学模型，减小消除安时积分法存在的累计误差问题。对比传统安时积分法估算精度，结果表明：SOC拐点修正安时积分实时估算法的误差在3%，说明该方法在实际工况中具有可行性，并且估算精度较高，可为SOC实时估算与检测提供重要参考。     

SOC estimation of lithium battery based on adaptive untracked Kalman filter

电池荷电状态（SOC）的准确估计是电池管理系统的关键问题,对电池的可靠性和安全性至关重要。由于多数情况下建立的电池模型精度不够高、电池系统的噪声统计是未知的或不准确的,这都会对锂离子电池系统的SOC估计会产生较大影响。本文采用二阶RC等效模型,可减小电池模型带来的误差;同时结合SageHusa滤波算法与无迹卡尔曼滤波（UKF）算法提出了一种新的SOC估计方法,基于噪声统计估计器的自适应无迹卡尔曼（AUKF）滤波算法,它可以对系统噪声进行实时修正以提高SOC的估算精度。并通过比较AUKF和UKF来验证SOC估计方法的准确性和有效性。实验结果表明,AUKF具有更高的SOC估计精度和自适应能力,在脉冲放电工况和动态工况下的估计精度均能保持在4.68%以内,可以有效地估计电池的SOC值。     

基于蚁群算法的神经网络绝缘子闪络电压预测

为提高覆冰绝缘子闪络电压的预测精度及预测速度,采用BP神经网络和蚁群算法相结合的方法进行预测模型设计。利用闪络电压及其影响因素之间的试验数据,建立其神经网络的预测模型。以网络的权值和阈值为自变量,通过蚁群算法的迭代运算,搜索出误差全局最小值,再进行网络的二次学习训练。结果表明,该方法具有较高的预测精度,适用于绝缘子闪络电压的预测。     

The study of control strategy for urban mass transit based on flywheel energy storage system

针对城市轨道交通列车运行密度高,起制动功率大的特点,采用飞轮型再生制动能量回收装置可有效降低直流牵引网压波动,降低牵引能耗。由于该装置采用基于直流牵引网母线电压高低进行充放电的控制策略,在实际运行工况中可能存在无法准确识别再生能或储能设备SOC值无法自动调整导致无法再响应牵引网压波动的情况,本文提出空载网压识别和SOC自适应控制策略进行解决,通过轨道交通试验平台的试验验证,得出该控制策略的有效性。