WO2022100229A1 - 一种基于人工智能的锂离子电池系统soc估计方法 - Google Patents

一种基于人工智能的锂离子电池系统soc估计方法 Download PDF

Info

Publication number
WO2022100229A1
WO2022100229A1 PCT/CN2021/116032 CN2021116032W WO2022100229A1 WO 2022100229 A1 WO2022100229 A1 WO 2022100229A1 CN 2021116032 W CN2021116032 W CN 2021116032W WO 2022100229 A1 WO2022100229 A1 WO 2022100229A1
Authority
WO
WIPO (PCT)
Prior art keywords
charging
soc
segment
battery
deep learning
Prior art date
Application number
PCT/CN2021/116032
Other languages
English (en)
French (fr)
Inventor
熊瑞
田金鹏
段砚州
Original Assignee
北京理工大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京理工大学 filed Critical 北京理工大学
Priority to US17/916,048 priority Critical patent/US11796597B2/en
Publication of WO2022100229A1 publication Critical patent/WO2022100229A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/385—Arrangements for measuring battery or accumulator variables
    • G01R31/387—Determining ampere-hour charge capacity or SoC
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • G01R31/3828—Arrangements for monitoring battery or accumulator variables, e.g. SoC using current integration

Definitions

  • the present invention relates to the field of battery systems, and in particular, to estimation of the state of charge of a lithium-ion battery system.
  • the battery management system can only obtain signals such as voltage, current, and temperature, while the state of charge (SOC) cannot be directly measured, and needs to be estimated based on the sampling signal.
  • the battery SOC estimation method mainly uses the battery model combined with the ampere-hour integral method to estimate the SOC during the discharge process. Such methods rely on the relationship between SOC and voltage modeling, and use ampere-hour integration for closed-loop correction, which has some limitations in implementation.
  • the voltage is not sensitive to SOC changes, so the SOC estimation effect is poor.
  • the battery voltage model reduces the voltage simulation accuracy at low temperature, high current, and low SOC.
  • the dynamic discharge process of the battery is changeable.
  • the discharge condition of the vehicle power battery depends on the driver's habits, region, season, weather and other factors. Under these different conditions, it is difficult for the algorithm developed based on only a few offline experiments to guarantee the robustness of its large-scale application. Therefore, there is still a lack of a method for estimating the state of charge of a lithium-ion battery system with good adaptability for complex and changeable actual use conditions, a relatively simplified execution process, and high robustness in the art.
  • the present invention provides an artificial intelligence-based SOC estimation method for a lithium-ion battery system, which specifically includes the following steps:
  • Step 1 Obtain the daily charging curve of the battery system using various charging methods as training data
  • Step 2 Divide the charging curve into data segments, and calibrate the SOC of the last point of the data segment
  • Step 3 selecting an applicable deep learning algorithm, using each data segment obtained in step 2 to train the algorithm, and establishing a mapping relationship between each data segment and the last SOC of the segment;
  • Step 4 Practically apply the deep learning algorithm trained in Step 3, take the charging segment data collected by the battery management system as the input of the deep learning algorithm, and output the estimated battery SOC;
  • Step 5 Between every two charging processes, use the ampere-hour integration algorithm to perform the recursive calculation of SOC.
  • the deep learning algorithm is retrained and updated using the charging curve collected by the battery management system.
  • step 1 common charging methods including constant current charging, constant current and constant voltage charging, multi-stage constant current charging, pulse charging, etc. can be adopted when obtaining the daily charging curve; the obtained curve includes battery charging current, Voltage, temperature and other parameters; through the ampere-hour integration method, the capacity of the battery is obtained, and the SOC at each moment on the charging curve is calculated.
  • dividing the charging curve into data segments in the second step specifically includes: determining a preset segment length, and sliding the preset segment length on the charging curve, thereby dividing the charging curve obtained in step 1 into a plurality of segments.
  • a data segment with a preset segment length, each segmented segment includes a sampling signal sequence at each moment, such as voltage, current, temperature, etc.; at the same time, the SOC of the last point of each segment is determined.
  • the deep learning algorithm in the third step selects convolutional neural networks, densely connected networks, cyclic neural networks, etc., and uses the preferred gradient descent algorithm and its various variants to train it;
  • the data segment is used as the input of the deep learning algorithm, and the last point SOC corresponding to each segment is used as the output of the algorithm.
  • the ampere-hour integration method is used to calculate the SOC change value between two charging processes, which is used for the recurrence of the SOC.
  • the method provided by the present invention makes full use of the controllable charging process, uses a deep learning algorithm to train the relationship between the charging signal segment and the SOC, collects the signal segment in each charging process to calibrate the battery SOC, and charges twice.
  • the SOC estimation between processes is realized by ampere-hour integration, which avoids the negative impact of poor modeling effect and high uncertainty of the discharge process on the battery SOC estimation.
  • the deep learning algorithm is retrained by using the re-collected charging curve data after full charging and full discharging, so that the method of the present invention has the characteristics of self-adaptation, and can realize SOC estimation through migration learning when the working conditions of the battery change. A quick update of the algorithm.
  • Fig. 1 is the flow chart of the method provided by the present invention.
  • FIG. 2 is a schematic diagram of the process of SOC estimation for a certain lithium iron phosphate battery based on the present invention
  • FIG. 3 is the result of SOC estimation of a certain lithium iron phosphate battery based on the present invention.
  • FIG. 1 The artificial intelligence-based SOC estimation method for a lithium-ion battery system provided by the present invention is shown in FIG. 1, and specifically includes the following parts:
  • Step 1 Use common charging schemes such as constant current charging, constant current and constant voltage charging, multi-stage constant current charging, and pulse charging to charge the battery, and use the battery test and battery management system to obtain the daily charging curve of the battery system as training. data. Through the ampere-hour integration method, the capacity of the battery is obtained, and the SOC at each moment of the charging curve is calculated.
  • Step 2 Divide the charging curve into data segments, and demarcate the SOC of the last point of the data segment, which specifically includes: determining a preset segment length, sliding the preset segment length on the charging curve, and then using the charging data obtained in step 1.
  • the curve is divided into data segments of this length, and each segment includes a sequence of sampled signals at each moment, such as voltage, current, temperature, and so on. At the same time, for each segment the SOC of its last point is determined.
  • Step 3 Use deep learning algorithms such as convolutional neural networks, densely connected networks, and recurrent neural networks to establish a mapping relationship between the data segment and the last point SOC of the segment.
  • the input of the deep learning algorithm is the data segment in step 2, and the output is the last SOC corresponding to the segment.
  • the learning algorithm is trained using the gradient descent algorithm and its various variants.
  • Step 4 In the actual application process of the battery system, collect the charging segment data as the input of the deep learning algorithm, and output the battery SOC, which specifically includes: when the battery is actually running, the battery management system collects data based on the settings in step 2 during the charging process. Charge the segment, and use the collected charging segment as the input of the deep learning algorithm trained in step 3, thereby outputting the SOC corresponding to the last point of the segment.
  • the battery management system collects data based on the settings in step 2 during the charging process. Charge the segment, and use the collected charging segment as the input of the deep learning algorithm trained in step 3, thereby outputting the SOC corresponding to the last point of the segment.
  • the battery management system collects data based on the settings in step 2 during the charging process. Charge the segment, and use the collected charging segment as the input of the deep learning algorithm trained in step 3, thereby outputting the SOC corresponding to the last point of the segment.
  • a deep recurrent neural network is used for training.
  • the 10-minute voltage and current data segments collected by the battery management system can be used to estimate the SOC at the last point of the 10-minute segment.
  • the algorithm can be made to handle data fragments of arbitrary length by marking parts of the fragment data as missing (eg, set to 0) during training. For example, when the collected data segment is less than 10 minutes, the corresponding missing flag (such as 0) can be filled in the data, and the filling is 10 minutes, which can be used as the input of the algorithm to realize SOC estimation.
  • Step 5 Between two charging processes, use the ampere-hour integration algorithm to calculate the SOC recursively, which specifically includes: taking the SOC at the last point of the segment estimated in step 4 as the initial value, and running the ampere-hour integration to perform the SOC recursion Calculation, as shown in process 3 in Figure 2, the ampere-hour integration calculates the change value of SOC by calculating the battery charge and discharge amount, so that the SOC recursion between two charging processes can be realized. Until an available charging segment is collected during the next charging process of the battery, return to step 4 to update the battery SOC.
  • the SOC estimation result and SOC estimation error of the lithium iron phosphate battery in the charging process in this embodiment are shown in FIG. 3 .
  • the charging curve is divided into consistent charging curve segments in step 2, and the SOC of the corresponding last point is calculated as new training data for updating the deep learning algorithm.
  • the SOC estimation algorithm can be updated by fine-tuning some parameters in the pre-trained deep learning algorithm.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Secondary Cells (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)

Abstract

一种基于人工智能的锂离子电池系统荷电状态(State of Charge，SOC)估计方法。该方法通过深度学习手段建立电池系统充电片段数据与荷电状态之间的关系，能够实现在充电过程的任意阶段对荷电状态进行校正。放电过程的荷电状态估计则采用安时积分进行。所提出的估计方法可以随电池系统工作状态变化自适应更新。

Description

一种基于人工智能的锂离子电池系统SOC估计方法 技术领域
本发明涉及电池系统领域，尤其涉及锂离子电池系统的荷电状态估计。
背景技术
在锂离子电池的运行过程中，电池管理系统仅能获得电压、电流、温度等信号，而对于荷电状态(State of charge,SOC)则不能直接测量，需要基于采样信号进行估计。目前电池SOC估计的方法主要使用电池模型结合安时积的分方法，在放电过程进行SOC估计。此类方法依赖于SOC与电压建模之间的关系，利用安时积分用于实现闭环校正，在实施过程中存在一些局限性。首先，对于某些电池体系，如磷酸铁锂或钛酸锂电池，其电压对SOC变化不敏感，因此SOC估计效果较差。此外，电池电压模型在低温、大电流、低SOC时电压仿真精度降低。另外，电池的动态放电过程多变，如车用动力电池的放电工况依赖于驾驶员习惯、地区、季节、天气等因素。这些不同的情况下，仅基于少量离线试验开发的算法难以保证其大范围应用时的鲁棒性。因此，本领域中尚缺乏针对复杂多变的实际使用情况具有较好的自适应性，且执行过程相对简化、鲁棒性高的锂离子电池系统荷电状态估计方法。
发明内容
有鉴于此，本发明提供了一种基于人工智能的锂离子电池系统SOC估计方法，具体包括以下步骤：
步骤一、获取电池系统采用多种充电方式的日常充电曲线作为训练数据；
步骤二、将充电曲线分割为数据片段，并标定数据片段最后一点的SOC；
步骤三、选择适用的深度学习算法，利用由步骤二获取的各数据片段对所述算法进行训练，建立每个数据片段与该片段最后一点SOC间的映射关系；
步骤四、将经过步骤三训练好的深度学习算法实际应用，以电池管理系统采集的充电片段数据作为所述深度学习算法的输入，输出估计的电池SOC；
步骤五、在每两次充电过程之间，使用安时积分算法进行SOC的递推计算。
进一步地，在锂离子电池系统经历过满充、满放等操作历程后，利用电池管理系统采集的充电曲线重新对所述深度学习算法进行训练更新。
进一步地，所述步骤一中在获取日常充电曲线时可采用包括恒流充电、恒流恒压充电、多阶恒流充电、脉冲充电等的常见充电方式；所获得的曲线包括电池充电电流、电压、温度等参数；通过安时积分方法，获得电池的容量，并计算充电曲线上各个时刻的SOC。
进一步地，所述步骤二中将充电曲线分割为数据片段具体包括：确定一个预设片段长度，将预设片段长度在充电曲线上滑动，由此将步骤一获取的充电曲线 划分为包含多个预设片段长度的数据片段，每个分割的片段包括各个时刻的采样信号序列，例如电压、电流、温度等；同时，对于每个片段确定其最后一点的SOC。
进一步地，所述步骤三中的深度学习算法选用卷积神经网络、密集连接网络、循环神经网络等，使用优选梯度下降算法及其各种变体对其进行训练；以步骤二中分割的各数据片段作为深度学习算法输入，并且以各片段对应的最后一点SOC作为算法输出。
进一步地，所述步骤五中利用安时积分法计算两次充电过程之间的SOC变化值，用于对SOC的递推。
上述本发明所提供的方法，充分利用充电过程可控的特点，使用深度学习算法训练充电信号片段与SOC之间的关系，在每次充电过程中采集信号片段进行电池SOC的标定，两次充电过程之间的SOC估计则由安时积分实现，由此避免了放电过程建模效果差、不确定性高对电池SOC估计的负面影响。此外，利用经历过满充、满放后重新采集的充电曲线数据对深度学习算法重新训练，使得本发明的方法具有自适应的特点，可以在电池工作条件发生变化时，通过迁移学习实现SOC估计算法的快速更新。
附图说明
图1是本发明所提供方法的流程图；
图2是基于本发明对某款磷酸铁锂电池进行SOC估计的过程示意图；
图3是基于本发明对某款磷酸铁锂电池进行SOC估计的结果。
具体实施方式
上述仅是本发明技术方案的概述，为了能够更清楚了解本发明的技术手段，以下结合附图与具体实施方式对本发明作进一步的详细说明。
本发明所提供的基于人工智能的锂离子电池系统SOC估计方法如附图1所示，具体包括以下部分：
步骤一、采用如恒流充电、恒流恒压充电、多阶恒流充电、脉冲充电等常见的充电方案，对电池进行充电，利用电池试验、电池管理系统获取电池系统的日常充电曲线作为训练数据。通过安时积分方法，获得电池的容量，并计算充电曲线各个时刻的SOC。
步骤二、将充电曲线分割为数据片段，并标定数据片段最后一点的SOC，具体包括：确定一个预设片段长度，将预设片段长度在充电曲线上滑动，由此将步骤一中获取的充电曲线划分为该长度的数据片段，每个片段包括各个时刻的采样信号序列，例如电压、电流、温度等。同时，对于每个片段确定其最后一点的SOC。
步骤三、使用卷积神经网络、密集连接网络、循环神经网络等深度学习算法建立数据片段与片段最后一点SOC间的映射关系。深度学习算法的输入是步骤二中的数据片段，输出则是片段对应的最后一点SOC。采用梯度下降算法及其各种变体对学习算法进行训练。
步骤四、在电池系统实际应用过程中，采集充电片段数据作为深度学习算法的输入，输出电池SOC，具体包括：在电池实际运行时，电池管理系统在充电过程中基于步骤二中的设定采集充电片段，并将采集到的充电片段作为步骤三中训练好的深度学习算法的输入，由此输出片段最后一点对应的SOC。在本发明的一个优选实施例中，如图2过程①②所示，针对某款磷酸铁锂电池的0.3C恒流恒压充电，将其划分为最长为10分钟长度的电压、电流数据片段。以该片段作为输入，片段最后一点的SOC作为输出，采用深度循环神经网络进行训练。训练后即可使用电池管理系统采集的10分钟长度的电压、电流数据片段估计10分钟片段最后一点的SOC。此外，通过在训练过程中将片段数据中的部分标记为缺失(如设置为0)，可以使算法处理任意长度的数据片段。例如，当采集到的数据片段不足10分钟时，可以为数据填充相应的缺失标记(如0)，补齐为10分钟，作为算法输入，实现SOC估计。
步骤五、在两次充电过程之间，使用安时积分算法进行SOC的递推计算，具体包括：将步骤四中估计的片段最后一点的SOC作为初始值，运行安时积分进行SOC的递推计算，如图2过程③所示，安时积分通过计算电池充放电量来计算SOC的变化值，由此可以实现两次充电过程之间的SOC递推。直至电池下一次充电过程采集到可用的充电片段，则返回步骤四，更新电池SOC。本实施例中磷酸铁锂电池在充电过程的SOC估计结果及SOC估计误差如图3所示。
步骤六、在电池系统经历满充、放光等操作后，采集相应充电曲线，对电池的SOC估计算法进行更新，具体包括：在电池系统使用过程中经历了充电至上截止电压(此时SOC＝100％)、放电至下截止电压(此时SOC＝0)等过程后，采集包含以上两个过程之一的充电曲线。将该充电曲线划分为步骤二中一致的充电曲线片段，并计算相应最后一点的SOC，作为新的训练数据，用于更新深度学习算法。可通过微调预先训练的深度学习算法中的部分参数更新SOC估计算法。
尽管已经示出和描述了本发明的实施例，对于本领域的普通技术人员而言，可以理解在不脱离本发明的原理和精神的情况下可以对这些实施例进行多种变化、修改、替换和变型，本发明的范围由所附权利要求及其等同物限定。

Claims (5)

  1. 一种基于人工智能的锂离子电池系统SOC估计方法，其特征在于：具体包括以下步骤：
    步骤一、获取电池系统采用多种充电方式的日常充电曲线作为训练数据；
    步骤二、将充电曲线分割为数据片段，并标定数据片段最后一点的SOC，具体包括：确定一个预设片段长度，将预设片段长度在充电曲线上滑动，由此将步骤一获取的充电曲线划分为包含多个预设片段长度的数据片段，每个分割的片段包括各个时刻的采样信号序列；同时，对于每个片段确定其最后一点的SOC；
    步骤三、选择适用的深度学习算法，利用由步骤二获取的各数据片段对所述算法进行训练，以步骤二中分割的各数据片段作为深度学习算法输入，并且以各片段对应的最后一点SOC作为算法输出，建立每个数据片段与该片段最后一点SOC间的映射关系；
    步骤四、将经过步骤三训练好的深度学习算法实际应用，以电池管理系统采集的充电片段数据作为所述深度学习算法的输入，输出估计的电池SOC；
    步骤五、在每两次充电过程之间，使用安时积分算法进行SOC的递推计算。
  2. 如权利要求1所述的方法，其特征在于：在锂离子电池经历过满充、满放操作历程后，利用电池管理系统采集的充电曲线重新对所述深度学习算法进行训练更新。
  3. 如权利要求1所述的方法，其特征在于：所述步骤一中在获取日常充电曲线时采用恒流充电、或者恒流恒压充电、或者多阶恒流充电、或者脉冲充电方式；所获得的曲线包括电池充电电流、和电压、和温度参数；通过安时积分方法，获得电池的容量，并计算充电曲线上各个时刻的SOC。
  4. 如权利要求1所述的方法，其特征在于：所述步骤三中的深度学习算法选用卷积神经网络、或者密集连接网络、或者循环神经网络，利用梯度下降算法及其各种变体对深度学习算法进行训练。
  5. 如权利要求1所述的方法，其特征在于：所述步骤五中利用安时积分法计算两次充电过程之间的SOC变化值，用于对SOC的递推。
PCT/CN2021/116032 2020-11-16 2021-09-01 一种基于人工智能的锂离子电池系统soc估计方法 WO2022100229A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/916,048 US11796597B2 (en) 2020-11-16 2021-09-01 Method for estimating state of charge (SOC) of lithium-ion battery system based on artificial intelligence (AI)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011280497.9 2020-11-16
CN202011280497.9A CN112379272B (zh) 2020-11-16 2020-11-16 一种基于人工智能的锂离子电池系统soc估计方法

Publications (1)

Publication Number Publication Date
WO2022100229A1 true WO2022100229A1 (zh) 2022-05-19

Family

ID=74584772

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/116032 WO2022100229A1 (zh) 2020-11-16 2021-09-01 一种基于人工智能的锂离子电池系统soc估计方法

Country Status (3)

Country Link
US (1) US11796597B2 (zh)
CN (1) CN112379272B (zh)
WO (1) WO2022100229A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112379272B (zh) 2020-11-16 2021-09-21 北京理工大学 一种基于人工智能的锂离子电池系统soc估计方法
CN114062948B (zh) * 2022-01-11 2022-05-20 南通大学 一种基于3dcnn的锂离子电池soc估计方法
CN115032541B (zh) * 2022-04-13 2023-12-05 北京理工大学 一种锂离子电池容量退化预测方法及装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107769335A (zh) * 2017-12-01 2018-03-06 深圳市森树强电子科技有限公司 一种多模式锂电池智能充电管理方法及装置
CN109031147A (zh) * 2018-08-21 2018-12-18 湖南兴业绿色电力科技有限公司 一种磷酸铁锂电池组的soc估算方法
CN109633470A (zh) * 2019-01-04 2019-04-16 深圳市计量质量检测研究院 基于ekf-gpr和日常片段数据的电池实时全充时间的估算方法
CN110058175A (zh) * 2019-05-05 2019-07-26 北京理工大学 一种动力电池开路电压-荷电状态函数关系的重构方法
CN110673039A (zh) * 2019-11-11 2020-01-10 安徽优旦科技有限公司 一种基于大数据的磷酸铁锂电池soc充电在线校正方法
CN112379272A (zh) * 2020-11-16 2021-02-19 北京理工大学 一种基于人工智能的锂离子电池系统soc估计方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1702219B1 (en) * 2003-12-18 2012-05-02 LG Chemical Co., Ltd. Apparatus and method for estimating state of charge of battery using neural network
CN105093114B (zh) * 2015-03-02 2017-12-08 江苏卓岸电源科技有限公司 一种电池在线建模与荷电状态的联合估计方法及系统
CN108519556A (zh) * 2018-04-13 2018-09-11 重庆邮电大学 一种基于循环神经网络的锂离子电池soc预测方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107769335A (zh) * 2017-12-01 2018-03-06 深圳市森树强电子科技有限公司 一种多模式锂电池智能充电管理方法及装置
CN109031147A (zh) * 2018-08-21 2018-12-18 湖南兴业绿色电力科技有限公司 一种磷酸铁锂电池组的soc估算方法
CN109633470A (zh) * 2019-01-04 2019-04-16 深圳市计量质量检测研究院 基于ekf-gpr和日常片段数据的电池实时全充时间的估算方法
CN110058175A (zh) * 2019-05-05 2019-07-26 北京理工大学 一种动力电池开路电压-荷电状态函数关系的重构方法
CN110673039A (zh) * 2019-11-11 2020-01-10 安徽优旦科技有限公司 一种基于大数据的磷酸铁锂电池soc充电在线校正方法
CN112379272A (zh) * 2020-11-16 2021-02-19 北京理工大学 一种基于人工智能的锂离子电池系统soc估计方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
CHAORAN LI, FEI XIAO, YAXIANG FAN, XIN TANG, GUORUN YANG: "Joint Estimation of the State of Charge and the State of Health Based on Deep Learning for Lithium-ion Batteries", PROCEEDINGS OF THE CSEE, ZHONGGUO DIANJI GONGCHENG XUEHUI, CN, vol. 41, no. 2, 10 April 2020 (2020-04-10), CN , pages 681 - 692, XP055929392, ISSN: 0258-8013, DOI: 10.13334/j.0258-8013.pcsee.191867 *
YUEJIU ZHENG ET AL.: "Investigating the error sources of the online state of charge estimation methods for lithium-ion batteries in electric vehicles", JOURNAL OF POWER SOURCES, 22 December 2017 (2017-12-22), XP085322261 *

Also Published As

Publication number Publication date
US20230204668A1 (en) 2023-06-29
US11796597B2 (en) 2023-10-24
CN112379272B (zh) 2021-09-21
CN112379272A (zh) 2021-02-19

Similar Documents

Publication Publication Date Title
WO2022100229A1 (zh) 一种基于人工智能的锂离子电池系统soc估计方法
CN112946499B (zh) 基于机器学习的锂电池健康状态及荷电状态联合估算方法
CN107368619B (zh) 一种扩展卡尔曼滤波soc估算方法
CN111965559B (zh) 一种锂离子电池soh在线估计方法
CN108693472A (zh) 电池等效模型参数在线辨识方法
CN110673039B (zh) 一种基于大数据的磷酸铁锂电池soc充电在线校正方法
CN111060824B (zh) 一种基于模型与算法分步融合的荷电状态估计方法
CN110133507B (zh) 一种基于narx-ukf算法的电池剩余电量估计方法
CN104569844A (zh) 阀控密封式铅酸蓄电池健康状态监测方法
CN110658459B (zh) 基于双向循环神经网络的锂离子电池荷电状态估计方法
CN109669131B (zh) 一种工况环境下动力电池soc估算方法
CN107688155B (zh) 一种用于电池管理系统中的电池剩余容量估算方法
CN107153163A (zh) 一种基于自适应ukf的锂电池soc估算方法
CN105699910A (zh) 一种锂电池剩余电量在线估计方法
CN107436412B (zh) 一种基于自学习估算动力电池功率方法
WO2022100230A1 (zh) 基于人工智能的锂离子电池充电曲线重构及状态估计方法
CN112649747A (zh) 一种分数阶扩展卡尔曼的锂电池soc估算方法
CN110441702B (zh) 一种用扩展卡尔曼滤波估算锂离子电池充电电量的方法
CN103278777A (zh) 一种基于动态贝叶斯网络的锂电池健康状况估计方法
WO2023077802A1 (zh) 一种电池容量特征提取方法
CN108693483A (zh) 一种电池包荷电状态的计算方法及系统
CN106154168A (zh) 数据驱动的动力电池荷电状态估计方法
CN112269133B (zh) 一种基于预充电路模型参数识别的soc估计方法
CN110716141B (zh) 一种基于扩展卡尔曼滤波的电池荷电状态估计方法
CN106443496A (zh) 一种带改进型噪声估计器的电池荷电状态估计方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21890757

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21890757

Country of ref document: EP

Kind code of ref document: A1