WO2024093005A1 - 一种电池容量预测方法、装置及电子设备 - Google Patents

一种电池容量预测方法、装置及电子设备 Download PDF

Info

Publication number
WO2024093005A1
WO2024093005A1 PCT/CN2022/143343 CN2022143343W WO2024093005A1 WO 2024093005 A1 WO2024093005 A1 WO 2024093005A1 CN 2022143343 W CN2022143343 W CN 2022143343W WO 2024093005 A1 WO2024093005 A1 WO 2024093005A1
Authority
WO
WIPO (PCT)
Prior art keywords
battery
tested
model
characteristic parameter
capacity
Prior art date
Application number
PCT/CN2022/143343
Other languages
English (en)
French (fr)
Inventor
张宇平
别传玉
刘虹灵
骆凡
陶君
朱传奇
王雪晴
王朝京
王远洋
宋华伟
Original Assignee
武汉动力电池再生技术有限公司
荆门动力电池再生技术有限公司
天津动力电池再生技术有限公司
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 武汉动力电池再生技术有限公司, 荆门动力电池再生技术有限公司, 天津动力电池再生技术有限公司 filed Critical 武汉动力电池再生技术有限公司
Publication of WO2024093005A1 publication Critical patent/WO2024093005A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367Software therefor, e.g. for battery testing using modelling or look-up tables

Definitions

  • the present invention relates to the field of battery detection, and in particular to a battery capacity prediction method, device and electronic equipment.
  • batteries have been widely used in the energy field and in all aspects of our lives.
  • batteries will gradually age during use and their capacity will gradually decrease, thus affecting the normal use of the batteries.
  • the current battery capacity prediction methods are roughly divided into traditional methods and machine learning methods.
  • the traditional method generally predicts the battery capacity by combining the corresponding formula after a period of charging and discharging;
  • the machine learning method generally predicts through support vector machines, Gaussian regression processes, deep learning networks, etc., and also has good prediction results.
  • the present invention provides a battery capacity prediction method, comprising:
  • the second characteristic parameter is used as the input of the capacity prediction model, and the estimated capacity of the battery to be tested is output and stored in the sample database.
  • the obtaining of basic information of the battery to be tested includes:
  • the scanning device is controlled to scan and collect basic information of the battery to be tested, wherein the basic information includes serial number information, type information and nominal capacity information.
  • determining a feature extraction model, a feature noise reduction model, and a capacity prediction model of the battery to be tested according to the basic information includes:
  • the corresponding feature extraction model, feature noise reduction model and capacity prediction model are determined according to the serial number information, type information and nominal capacity information of the battery to be tested and the mapping relationship.
  • test data of the battery under test for charge and discharge test including:
  • the test data is obtained according to the battery excitation response signal.
  • test data is effectively intercepted based on the feature extraction model to obtain the first characteristic parameter of the battery to be tested, including:
  • the test data in the pulse charge and discharge process is intercepted, and the test data is segmented to obtain multiple segment voltage values, and multiple static voltage values corresponding to the multiple segment voltage values are extracted after the battery pulse test is completed and the battery is static.
  • the charge and discharge ohmic internal resistance and the charge and discharge polarization internal resistance are calculated based on the segment voltage values and the static voltage values to obtain the pulse charge and discharge test data;
  • the test data of the short-time charging process is intercepted to determine the charging time, platform voltage, open circuit voltage, current value and start and end voltage values after standing, and the charging capacity is calculated according to the charging time and current value, and the rebound voltage is calculated according to the start and end voltage values to obtain the short-time charging test data;
  • the first characteristic parameter of the battery to be tested is obtained according to the pulse charge and discharge test data and the short-time charge test data.
  • the first characteristic parameters include: charge and discharge ohmic internal resistance, charge and discharge polarization internal resistance, open circuit voltage, charge capacity, platform voltage, charge time and rebound voltage.
  • performing denoising processing on the first characteristic parameter based on the characteristic denoising model to obtain a second characteristic parameter includes:
  • the first characteristic parameter of the battery to be tested is used as input to obtain a correlation coefficient of the characteristic parameter, and the first characteristic parameter is reduced according to the correlation coefficient to obtain a second characteristic parameter of the battery to be tested.
  • the second characteristic parameter is used as the input of the capacity prediction model, and the estimated capacity of the battery to be tested is output and stored in the sample database, including:
  • the capacity prediction model is obtained based on a neural network, a support vector machine, or deep learning training;
  • the basic information, the second characteristic parameter and the estimated capacity of the battery to be tested are stored in a sample database.
  • the present invention also provides a battery capacity prediction device, comprising:
  • An information acquisition unit used to acquire basic information of the battery to be tested
  • a model determination unit used to determine a feature extraction model, a feature noise reduction model and a capacity prediction model of the battery to be tested according to the basic information
  • a battery testing unit used to obtain test data of a battery to be tested for charge and discharge testing
  • a data interception unit used for effectively intercepting the test data based on a feature extraction model to obtain a first characteristic parameter of the battery to be tested
  • a parameter denoising unit configured to perform denoising processing on the first characteristic parameter based on a characteristic denoising model to obtain a second characteristic parameter
  • the capacity estimation unit is used to use the second characteristic parameter as the input of the capacity prediction model, output the estimated capacity value of the battery to be tested, and store it in the sample database.
  • the present invention also provides an electronic device, including a memory and a processor.
  • the memory is used to store programs
  • the processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the battery capacity prediction method described in any one of the above implementations.
  • the beneficial effect of adopting the above embodiment is: the battery capacity prediction method provided by the present invention, by establishing multiple feature extraction models, feature noise reduction models, and capacity prediction models, calls different models to process the test data according to different battery types, and determines the predicted value of the capacity of the battery to be tested. Compared with the prior art that only predicts specific types of batteries, it can be compatible with the capacity prediction function of multiple batteries while ensuring accuracy.
  • FIG1 is a schematic flow chart of an embodiment of a battery capacity prediction method provided by the present invention.
  • FIG2 is a schematic flow chart of an embodiment of S102 in FIG1 of the present invention.
  • FIG3 is a schematic diagram of an embodiment of the process of S104 in FIG1 of the present invention.
  • FIG4 is a schematic diagram of an embodiment of capturing pulse charge and discharge test data provided by the present invention.
  • FIG5 is a schematic diagram of an embodiment of capturing short-time charging test data provided by the present invention.
  • FIG6 is a schematic flow chart of an embodiment of S105 in FIG1 of the present invention.
  • FIG. 7 is a schematic diagram of a flow chart of an embodiment of S106 in FIG. 1 of the present invention.
  • FIG8 is a schematic structural diagram of an embodiment of a battery capacity prediction device provided by the present invention.
  • FIG. 9 is a schematic diagram of the structure of an embodiment of an electronic device provided by the present invention.
  • the embodiments of the present invention provide a battery capacity prediction method, device and electronic device, which are described below respectively.
  • FIG1 is a schematic flow chart of an embodiment of a battery capacity prediction method provided by the present invention. As shown in FIG1 , the battery capacity prediction method includes:
  • S106 Use the second characteristic parameter as input of the capacity prediction model, output an estimated capacity value of the battery to be tested, and store it in a sample database.
  • the battery capacity prediction method provided in the embodiment of the present invention by establishing a variety of feature extraction models, feature noise reduction models and capacity prediction models, uses different models to predict the capacity of the battery to be tested according to different matches of the basic information of the battery, can effectively be compatible with various types of batteries, and can make accurate predictions for various types of batteries.
  • obtaining basic information of the battery to be tested includes:
  • the scanning device is controlled to scan and collect basic information of the battery to be tested, wherein the basic information includes serial number information, type information and nominal capacity information.
  • the scanning device can read the electronic tag of the battery by using RFID (Radio Frequency Identification) or by scanning the QR code tag of the battery through a QR code scanner to obtain the basic information of the battery, including numbering information, type information, and nominal capacity information; in a specific embodiment, the numbering information can be a first type battery, a second type battery, and a third type battery, the type information can be a ternary battery and a lithium iron phosphate battery, and the nominal capacity information can be 10Ah, 12Ah, and 20Ah.
  • RFID Radio Frequency Identification
  • FIG2 a method for determining a feature extraction model, a feature noise reduction model, and a capacity prediction model of a battery to be tested according to basic information is shown in FIG2 and includes:
  • S202 Determine a corresponding feature extraction model, feature noise reduction model and capacity prediction model according to the serial number information, type information and nominal capacity information of the battery to be tested and the mapping relationship.
  • the empirical model does not analyze the mechanism of the actual process, but instead performs mathematical statistical analysis on the data related to the process obtained from the actual process, and summarizes the mathematical relationship between the parameters and variables of the process according to the principle of minimum error.
  • the mathematical expression obtained in this way is called an empirical model.
  • the empirical model only considers input and output and is independent of the process mechanism. For example, when the battery is a first-class battery, through pre-training and testing, it is found that when the feature extraction model, feature denoising model and capacity prediction model are feature extraction model 1, feature denoising model 1 and capacity prediction model 1 respectively, the prediction result has the highest accuracy, that is, the mapping relationship is obtained: the corresponding models for the first-class battery are feature extraction model 1, feature denoising model 1 and capacity prediction model 1.
  • obtaining test data of a battery under test for charge and discharge testing includes:
  • the test data is obtained according to the battery excitation response signal.
  • the excitation signal refers to the input signal for testing the battery to be tested
  • the response signal refers to the output signal of the battery to be tested during the test process
  • the excitation signal can be pulse charging and discharging and short-time charging and discharging performed by a charging and discharging cabinet
  • the response signal can be the voltage and current data of the battery to be tested during the test obtained by the tester.
  • step S103 may be specifically as follows: setting a test step at the charging and discharging cabinet end, for example, for lithium iron phosphate batteries, adjusting the batteries to the same state, and performing pulse and short-time charging and discharging tests at a preset current (1C to 3C); based on the test step, controlling the charging and discharging equipment to generate an excitation signal, and acquiring the excitation response signal of the battery to be tested to obtain test data.
  • the process of effectively intercepting the test data based on the feature extraction model to obtain the first characteristic parameter of the battery to be tested is shown in FIG3 and includes:
  • pulse charge and discharge test data interception based on the feature extraction model, the test data in the pulse charge and discharge process is intercepted, the test data is segmented to obtain multiple segment voltage values, and multiple static voltage values corresponding to the multiple segment voltage values are extracted after the battery pulse test is completed and the static voltage values are used to calculate the charge and discharge ohmic internal resistance and the charge and discharge polarization internal resistance to obtain the pulse charge and discharge test data;
  • short-time charging test data interception intercept the test data in the short-time charging process based on the feature extraction model, determine the charging time, platform voltage, open circuit voltage, current value and start and end voltage values after standing, and calculate the charging capacity according to the charging time and current value, and calculate the rebound voltage according to the start and end voltage values to obtain the short-time charging test data;
  • S303 sorting out the first characteristic parameter: obtaining the first characteristic parameter of the battery to be tested according to the pulse charge and discharge test data and the short-time charge test data.
  • the method of calculating the ohmic internal resistance, polarization internal resistance, discharge capacity, and rebound voltage in the first characteristic parameter by intercepting the test data is shown in Figures 4 and 5, wherein b, c, f, and g in Figure 4 are pulse charge and discharge test process nodes, a, d, e, and h are corresponding static nodes, and the size of I is the same as the preset current size.
  • the preset current size can be 1C
  • t1 in Figure 5 is the start time point of short-time charging
  • t2 is the end time point of short-time charging
  • Ui and Uj are the start and end voltage values before and after standing for a period of time
  • the calculation method is:
  • Discharge ohmic internal resistance [(U a -U b ) + (U d -U c )] / 2I
  • the first characteristic parameter is subjected to denoising based on the characteristic denoising model to obtain the second characteristic parameter, as shown in FIG6 :
  • S602 Perform feature reduction on the first feature parameter according to the correlation coefficient to obtain a second feature parameter of the battery to be tested.
  • the process of denoising the first characteristic parameter can be: using the Pearson correlation coefficient or the grey correlation method to calculate the correlation of the first characteristic parameter, and reducing the first characteristic parameter by PCA (principal components analysis, also known as principal component analysis technology, which aims to use the idea of dimensionality reduction to transform multiple indicators into a few comprehensive indicators), or RFE (recursive feature elimination, which works on the principle of searching for feature subsets starting from all features in the training data set and successfully deleting features until the required number is retained), or LASSO (Least absolute shrinkage and selection operator, regression model, which is a compression estimation method that obtains a more refined model by constructing a penalty function) to obtain the second characteristic parameter.
  • PCA principal components analysis, also known as principal component analysis technology, which aims to use the idea of dimensionality reduction to transform multiple indicators into a few comprehensive indicators
  • RFE recursive feature elimination, which works on the principle of searching for feature subsets starting from all features in the training data set and successfully deleting features until the required number is retained
  • the second characteristic parameter is used as the input of the capacity prediction model, and the estimated capacity value of the battery to be tested is obtained as output and stored in the sample database, as shown in FIG7 :
  • S703 storing the basic information, the second characteristic parameter and the estimated capacity of the battery to be tested into a sample database.
  • S701 may be capacity prediction model 1, capacity prediction model 2 and capacity prediction model 3 trained by various methods such as neural network, support vector machine and deep learning;
  • S702 may be inputting the second characteristic parameter of the battery to be tested of the first type of battery into the capacity prediction model 1 to obtain the estimated capacity value of the battery to be tested;
  • S703 may be storing the basic information of the battery to be tested, the second characteristic parameter of the battery to be tested and the estimated capacity value of the battery to be tested into the sample database of the first type of battery, and improving the accuracy of the capacity prediction model by updating the sample database and regularly training the corresponding model.
  • S701 for newly added types of batteries to be tested, according to the basic information of the batteries to be tested, for newly added types of batteries to be tested with different type information, S701 can be a newly added corresponding capacity prediction model, and then the corresponding capacity prediction model is obtained by training through various methods such as neural networks, support vector machines and deep learning.
  • the corresponding test steps, characteristic parameter algorithms and capacity prediction models are different for different types of batteries. Batteries of the same type but different nominal capacities use the same capacity prediction model algorithm, but different training sets, and the predicted battery data can automatically update the training set. After adding different types of batteries, you can add a new capacity prediction model, and after adding batteries of the same type but different nominal capacities, you can add a new training set.
  • the embodiment of the present invention also provides a battery capacity prediction device.
  • the battery capacity prediction device 800 includes:
  • the information acquisition unit 801 is used to acquire basic information of the battery to be tested;
  • a model determination unit 802 is used to determine a feature extraction model, a feature noise reduction model and a capacity prediction model of the battery to be tested according to the basic information;
  • the battery testing unit 803 is used to obtain test data of the battery to be tested for charge and discharge testing
  • a data interception unit 804 is used to effectively intercept the test data based on a feature extraction model to obtain a first characteristic parameter of the battery to be tested;
  • a parameter denoising unit 805, configured to perform denoising on the first characteristic parameter based on a characteristic denoising model to obtain a second characteristic parameter;
  • the capacity prediction unit 806 is used to use the second characteristic parameter as the input of the capacity prediction model, output the estimated capacity value of the battery to be tested, and store it in the sample database.
  • the battery capacity prediction device 800 provided in the above embodiment can implement the technical solution described in the above battery capacity prediction method embodiment.
  • the specific implementation principles of the above modules or units can refer to the corresponding contents in the above battery capacity prediction method embodiment, which will not be repeated here.
  • the present invention also provides an electronic device 900.
  • the electronic device 900 includes a memory 901, a processor 902, an actuator 903 and a display 904.
  • FIG9 shows only some components of the electronic device 900, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.
  • the memory 901 may be an internal storage unit of the electronic device 900, such as a hard disk or memory of the electronic device 900. In other embodiments, the memory 901 may also be an external storage device of the electronic device 900, such as a mobile hard disk, a smart memory card, a flash memory card, etc. configured on the electronic device 900. In addition, the memory may also be a cloud storage device, such as a cloud hard disk and distributed storage, etc. The memory 901 is used to store the corresponding programs and data of the battery capacity prediction system.
  • the processor 902 may be a central processing unit (CPU), a microprocessor or other data processing chip, which is used to run the program code stored in the memory 901 or process data.
  • the processor 902 may also be a server or a server group.
  • the processor 902 may also be deployed on a cloud platform such as a cloud server, a cloud community or distributed cloud computing.
  • the display 903 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc.
  • the display 903 is used to display information on the electronic device 900 and to display a visual user interface.
  • the actuator 904 may be an RFID reader, a barcode scanner, a QR code scanner, a charging and discharging cabinet, and a single-chip computer group, etc.
  • the actuator 904 is used to receive signals from the processor 902, perform information reading, battery testing, and other steps, and feed back data to the processor 902.

Landscapes

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

Abstract

一种电池容量预测方法、装置及电子设备,其方法包括:获取待测电池的基本信息(S101);根据基本信息确定待测电池的特征提取模型、特征降噪模型以及容量预测模型(S102);获取待测电池进行充放电测试的测试数据(S103);基于特征提取模型对测试数据进行有效截取,得到待测电池的第一特征参数(S104);基于特征降噪模型对第一特征参数进行降噪处理,得到第二特征参数(S105);以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库(S106)。通过根据电池种类不同调用不同模型对测试数据进行处理,确定待测电池容量预测值,相比现有技术仅对特定种类的电池进行预测,可以在保证准确度的情况下兼容多种电池的容量预测功能。

Description

一种电池容量预测方法、装置及电子设备 技术领域
本发明涉及电池检测领域,具体涉及一种电池容量预测方法、装置及电子设备。
背景技术
随着电池技术的不断发展,电池在能源领域应用有着广泛的应用,遍布了我们生活的方方面面,而电池在使用过程中会逐渐老化,其容量也会逐渐降低,从而影响到电池的正常使用。
目前的电池容量预测方法大致分为传统的方法和机器学习方法,传统方法一般是通过一段时间充放电后结合对应公式来预测电池容量;机器学习方法一般通过支持向量机,高斯回归过程、深度学习网络等进行预测,也都有不错的预测效果。
但是目前的预测方法无论是传统方法还是机器学习方法,往往都只能预测特定种类电池的容量,而对于其他种类的电池无能为力。因此需要提出一种方法,可以兼容多种类电池并起到准确预测效果。
发明内容
有鉴于此,有必要提供一种电池容量预测方法及装置,用以实现对多种类型电池进行容量的准确预测。
为了实现上述目的,一方面,本发明提供了一种电池容量预测方法,包括:
获取待测电池的基本信息;
根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型;
获取待测电池进行充放电测试的测试数据;
基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数;
基于特征降噪模型对所述第一特征参数进行降噪处理,得到第二特征参数;
以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
进一步的,所述获取待测电池的基本信息,包括:
控制扫描设备扫描采集待测电池的基本信息,其中,所述基本信息包括编号信息、类型信息以及标称容量信息。
进一步的,根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型,包括:
基于预设的经验模型确定基本信息与特征提取模型、特征降噪模型以及容量预测模型的映射关系;
根据所述待测电池的编号信息、类型信息以及标称容量信息,以及所述映射关系确定对应的特征提取模型、特征降噪模型和容量预测模型。
进一步的,获取待测电池进行充放电测试的测试数据,包括:
基于充放电测试参数控制充放电设备生成一激励信号,以使待测电池生成激励响应信号;
根据电池激励响应信号得到所述测试数据。
进一步的,基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数,包括:
基于特征提取模型截取脉冲充放电过程中的测试数据,对测试数据进行分节处理得到多个分节电压值,并提取电池脉冲测试完成静置后与所述多个分节电压值对应的多个静置电压值,基于所述分节电压值和静置电压值计算充放电欧姆内阻和充放电极化内阻,得到脉冲充放电测试数据;
基于特征提取模型截取短时间充电过程中的测试数据,确定充电时间、平 台电压、开路电压、电流值和静置后的启止电压值,并根据充电时间和电流值计算充电电量,根据启止电压值计算回弹电压,得到短时间充电测试数据;
根据脉冲充放电测试数据、短时间充电测试数据,得到所述待测电池第一特征参数。
进一步的,第一特征参数包括:充放电欧姆内阻、充放电极化内阻、开路电压、充电容量、平台电压、充电时间及回弹电压。
进一步的,基于特征降噪模型对所述第一特征参数进行降噪处理,得到第二特征参数,包括:
基于所述降噪模型以所述待测电池的第一特征参数为输入得到特征参数的相关性系数,根据相关性系数对所述第一特征参数进行特征缩减,得到待测电池的第二特征参数。
进一步的,以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库,包括:
基于神经网络、或支持向量机、或深度学习训练得到所述容量预测模型;
基于容量预测模型,以所述待测电池的第二特征参数为输入,得到待测电池的容量预估值;
将待测电池的基本信息、第二特征参数和容量预估值存入样本数据库。
另一方面,本发明还提供了一种电池容量预测装置,包括:
信息获取单元,用于获取待测电池的基本信息;
模型确定单元,用于根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型;
电池测试单元,用于获取待测电池进行充放电测试的测试数据;
数据截取单元,用于基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数;
参数降噪单元,用于基于特征降噪模型对所述第一特征参数进行降噪处理, 得到第二特征参数;
容量预估单元,用于以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
另一方面,本发明还提供了一种电子设备,包括存储器和处理器。其中,
所述存储器,用于存储程序;
所述处理器,与所述存储器耦合,用于执行所述存储器中存储的所述程序,以实现上述任意一种实现方式中所述的电池容量预测方法中的步骤。
采用上述实施例的有益效果是:本发明提供的电池容量预测方法,通过建立多种特征提取模型、特征降噪模型、容量预测模型,根据电池种类不同调用不同的模型对测试数据进行处理,确定待测电池容量预测值,相比于现有技术中仅对特定种类的电池进行预测,可以在保证准确度的情况下兼容多种电池的容量预测功能。
附图说明
为了更清楚地说明本发明实施例中的技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1为本发明提供的电池容量预测方法的一个实施例流程示意图;
图2为本发明图1中S102的一个实施例流程示意图;
图3为本发明图1中S104的一个实施例流程示意图;
图4为本发明提供的截取脉冲充放电测试数据的一个实施例示意图;
图5为本发明提供的截取短时间充电测试数据的一个实施例示意图;
图6为本发明图1中S105的一个实施例流程示意图;
图7为本发明图1中S106的一个实施例流程示意图;
图8为本发明提供的电池容量预测装置的一个实施例结构示意图;
图9为本发明提供的电子设备的一个实施例结构示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述。显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域技术人员在没有做出创造性劳动前提下所获得的所以其他实施例,都属于本发明保护的范围。
应当理解,示意图的附图并未按实物比例绘制。本发明中使用的流程图示出了根据本发明的一些实施例实现的操作。应当理解,流程图的操作可以不按顺序实现,没有逻辑的上下文关系的步骤可以反转顺序或者同时实施。此外,本领域技术人员在本发明内容的指引下,可以向流程图添加一个或多个其他操作,也可以从流程图中移除一个或多个操作。
附图中所示的一些方框图是功能实体,不一定必须与物理或逻辑上独立的实体相对应。可以采用软件形式来实现这些功能实体,或在一个或多个硬件模块或集成电路中实现这些功能实体,或在不同网络和/或处理器系统和/或微控制器系统中实现这些功能实体。
在本文中提及“实施例”意味着,结合实施例描述的特定特征、结构或特性可以包含在本发明的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例,也不是与其它实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是,本文所描述的实施例可以与其它实施例相结合。
本发明实施例提供了一种电池容量预测方法、装置及电子设备,以下分别进行说明。
图1为本发明提供的电池容量预测方法的一个实施例流程示意图,如图1所示,电池容量预测方法包括:
S101、获取待测电池的基本信息;
S102、根据基本信息确定待测电池的特征提取模型、特征降噪模型以及容量预测模型;
S103、获取待测电池进行充放电测试的测试数据;
S104、基于特征提取模型对测试数据进行有效截取,得到待测电池的第一特征参数;
S105、基于特征降噪模型对第一特征参数进行降噪处理,得到第二特征参数;
S106、以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
与现有技术相比,本发明实施例提供的电池容量预测方法,通过建立多种特征提取模型、特征降噪模型及容量预测模型,根据电池基本信息的不同匹配使用不同的模型对待测电池进行容量预测,可以有效地兼容各类电池,并对各类电池均可做出准确预测。
在本发明的一些实施例中,获取待测电池的基本信息,包括:
控制扫描设备扫描采集待测电池的基本信息,其中,基本信息包括编号信息、类型信息以及标称容量信息。
具体的,扫描设备可以通过使用RFID(Radio Frequency Identification,射频识别技术)读取电池的电子标签,也可以通过二维码扫描器,扫描电池的二维码标签,从而获取电池基本信息,其包括编号信息、类型信息以及标称容量信息等;在具体的实施例中,编号信息可以是第一类电池、第二类电池和第三类电池,类型信息可以是三元电池和磷酸铁锂电池,标称容量信息可以是10Ah、12Ah、20Ah。
在本发明的一些实施例中,根据基本信息确定待测电池的特征提取模型、特征降噪模型以及容量预测模型的方法如图2所示,包括:
S201、基于预设的经验模型确定基本信息与特征提取模型、特征降噪模型 以及容量预测模型的映射关系;
S202、根据待测电池的编号信息、类型信息以及标称容量信息,以及映射关系确定对应的特征提取模型、特征降噪模型和容量预测模型。
需要说明的是,经验模型是指不分析实际过程的机理,而是根据从实际得到的与过程相关的数据进行数理统计分析、按误差最小原则,归纳出该过程各参数和变量之间得到数学关系式,用这种方法所得到的数学表达式称为经验模型,经验模型只考虑输入与输出而与过程机理无关,如:当电池为第一类电池时,通过预先训练测试得到当特征提取模型、特征降噪模型和容量预测模型分别是特征提取模型一、特征降噪模型一、容量预测模型一时,预测结果精确度最高,即得到映射关系为,第一类电池对应模型为特征提取模型一、特征降噪模型一、容量预测模型一。
在本发明的一些实施例中,获取待测电池进行充放电测试的测试数据,包括:
基于充放电测试参数控制充放电设备生成一激励信号,以使待测电池生成激励响应信号;
根据电池激励响应信号得到测试数据。
需要说明的是:激励信号是指对于待测电池进行测试的输入信号,相应信号是指测试过程待测电池的输出信号;具体的,在本发明的一个具体实施例中,激励信号可以是通过充放电柜进行的脉冲充放电和短时间充放电,响应信号可以是通过测试仪得到的待测电池测试过程电压和电流数据。
在本发明的一个具体实施例中,步骤S103可具体为:在充放电柜端设置测试工步,例如对磷酸铁锂电池,将电池调整至同一状态,对其在预设电流(1C~3C)下进行脉冲以及短时间充放电测试;基于测试工步,控制充放电设备生成一激励信号,采集设备获取待测电池的激励响应信号,得到测试数据。
在本发明的一些实施例中,基于特征提取模型对测试数据进行有效截取, 得到待测电池的第一特征参数的过程如图3所示,包括:
S301、脉冲充放电测试数据截取:基于特征提取模型截取脉冲充放电过程中的测试数据,对测试数据进行分节处理得到多个分节电压值,并提取电池脉冲测试完成静置后与多个分节电压值对应的多个静置电压值,基于分节电压值和静置电压值计算充放电欧姆内阻和充放电极化内阻,得到脉冲充放电测试数据;
S302、短时间充电测试数据截取:基于特征提取模型截取短时间充电过程中的测试数据,确定充电时间、平台电压、开路电压、电流值和静置后的启止电压值,并根据充电时间和电流值计算充电电量,根据启止电压值计算回弹电压,得到短时间充电测试数据;
S303、整理第一特征参数:根据脉冲充放电测试数据和短时间充电测试数据,得到待测电池第一特征参数。
在本发明的一个具体实施例中,具体地,通过截取的测试数据计算得到第一特征参数中欧姆内阻、极化内阻、放电容量、回弹电压的方法如图4、图5所示,其中图4中b、c、f、g是脉冲充放电测试过程分节点,a、d、e、h为对应静置节点,I大小与预设电流大小相同,具体的,预设电流大小可以为1C,图5中t 1是短时间充电开始时间点,t 2是短时间充电结束时间点,U i和U j是静置一段时间前后的启止电压值,计算方式为:
放电欧姆内阻=[(U a-U b)+(U d-U c)]/2I
放电极化内阻=(U b-U c)/I
充电欧姆内阻=[ (U f-U e)+(U g-U h)]/2I
充电极化内阻=(U g-U f)/I
时间t=t 2-t 1
放电容量
Figure PCTCN2022143343-appb-000001
回弹电压U=U i-U j
在本发明的一些实施例中,基于特征降噪模型对第一特征参数进行降噪处理,得到第二特征参数,如图6:
S601、基于降噪模型以待测电池的第一特征参数为输入得到特征参数的相关性系数;
S602、根据相关性系数对第一特征参数进行特征缩减,得到待测电池的第二特征参数。
通过对第一特征参数进行降噪处理得到第二特征参数,可以完成对第一特征参数的降维处理和对模型的优化,提升了电池容量预测系统的效率与预测精度;具体的,对第一特征参数进行降噪处理的过程可以是:使用Pearson相关系数或灰色关联度方式对第一特征参数进行相关性计算,通过PCA(principal components analysis,主成分分析技术,又称主分量分析技术,旨在利用降维的思想,把多指标转化为少数几个综合指标)、或RFE(recursive feature elimination,特征递归消除,工作原理是从训练数据集中的所有特征开始搜索特征子集,并成功地删除特征,直到保留所需的数量)、或LASSO(Least absolute shrinkage and selection operator,回归模型,是一种压缩估计方法,通过构造一个惩罚函数得到一个较为精炼的模型)方式对第一特征参数进行特征缩减,得到第二特征参数。
在本发明的一个具体实施例中,以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库,如图7:
S701、基于神经网络、或支持向量机、或深度学习训练得到容量预测模型;
S702、基于容量预测模型、以待测电池的第二特征参数为输入,得到待测电池的容量预估值;
S703、将待测电池的基本信息、第二特征参数和容量预估值存入样本数据库。
在本发明的一个具体实施例中,S701可以是通过神经网络、支持向量机和 深度学习等多种方式训练得到的容量预测模型一、容量预测模型二和容量预测模型三;S702可以是将第一类电池的待测电池一的第二特征参数输入容量预测模型一,得到待测电池一的容量预估值;S703可以是将待测电池一的基本信息、待测电池一的第二特征参数和待测电池一的容量预估值存入第一类电池的样本数据库中,通过更新样本数据库与定期训练对应模型,提升容量预测模型的精度。
在本发明的另一些具体实施例中,对于新增种类待测电池,根据待测电池的基本信息,对于类型信息不同的新增种类待测电池,S701可以是新增对应容量预测模型,再通过神经网络、支持向量机和深度学习等多种方式进行训练得到对应容量预测模型。
需要说明的是,对于容量预测系统,针对不同类型的电池,所对应的测试工步、特征参数算法以及容量预测模型也不同。类型相同标称容量不同的电池使用的容量预测模型算法相同,训练集不同,并且预测后的电池数据能自动对训练集进行更新。在新添不同类型的电池后,能够进行新增容量预测模型的操作,在新添同类型不同标称容量的电池后,能够进行新增训练集的操作。
为了更好实施本发明实施例中的电池容量预测方法,在电池容量预测方法基础之上,对应的,本发明实施例还提供了一种电池容量预测装置,如图8所示,电池容量预测装置800包括:
信息获取单元801,用于获取待测电池的基本信息;
模型确定单元802,用于根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型;
电池测试单元803,用于获取待测电池进行充放电测试的测试数据;
数据截取单元804,用于基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数;
参数降噪单元805,用于基于特征降噪模型对所述第一特征参数进行降噪 处理,得到第二特征参数;
容量预测单元806,用于以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
上述实施例提供的电池容量预测装置800可实现上述的电池容量预测方法实施例中描述的技术方案,上述各模块或单元具体实现的原理可参见上述电池容量预测方法实施例中的相应内容,此处不在赘述。
如图9所示,本发明还相应提供了一种电子设备900.该电子设备900包括存储器901、处理器902、执行器903及显示器904。图9仅示出了电子设备900的部分组件,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
存储器901在一些实施例中可以是电子设备900的内部存储单元,例如电子设备900的硬盘或内存。存储器901在另一些实施例中也可以是电子设备900的外部存储设备,例如电子设备900上配置的移动硬盘、智能存储卡、闪存卡等。此外,存储器也可以是云端存储设备,如云硬盘和分布式存储等。存储器901用于存储电池容量预测系统的对应程序和数据。
处理器902在一些实施例中可以是一中央处理器(Central Processing Unit,CPU)、微处理器或其他数据处理芯片,用于运行存储器901中存储的程序代码或处理数据。处理器902在另一些实施例中也可以是服务器或服务器组。处理器902也可以是部署在云端的云服务器、云社区或分布式云计算等云端平台上。
显示器903在一些实施例中可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode,有机发光二极管)触摸器等。显示器903用于显示在电子设备900的信息以及用于显示可视化的用户界面。
执行器904在一些实施例中可以是RFID读取器、扫码枪、二维码扫描器、充放电柜和单片机组等。执行器904用于接收处理器902信号,执行信息读取,电池测试等步骤,并反馈数据给处理器902。
本领域技术人员可以理解,实现上述实施例方法的全部或部分流程,可以通过计算机程序来指令相关的硬件来完成,所述的程序可存储于计算机可读存储介质中。其中,所述计算机可读存储介质为磁盘、光盘、只读存储记忆体或随机存储记忆体等。
以上对本发明所提供的电池容量预测方法、装置进行了详细介绍,本文中应用了具体个例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处,综上所述,本说明书内容不应理解为对本发明的限制。

Claims (10)

  1. 一种电池容量预测方法,其特征在于,包括:
    获取待测电池的基本信息;
    根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型;
    获取待测电池进行充放电测试的测试数据;
    基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数;
    基于特征降噪模型对所述第一特征参数进行降噪处理,得到第二特征参数;
    以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
  2. 根据权利要求1所述的电池容量预测方法,其特征在于,所述获取待测电池的基本信息,包括:
    控制扫描设备扫描采集待测电池的基本信息,其中,所述基本信息包括编号信息、类型信息以及标称容量信息。
  3. 根据权利要求2所述的电池容量预测方法,其特征在于,根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型,包括:
    基于预设的经验模型确定基本信息与特征提取模型、特征降噪模型以及容量预测模型的映射关系;
    根据所述待测电池的编号信息、类型信息以及标称容量信息,以及所述映射关系确定对应的特征提取模型、特征降噪模型和容量预测模型。
  4. 根据权利要求1所述的电池容量预测方法,其特征在于,获取待测电池进行充放电测试的测试数据,包括:
    基于充放电测试参数控制充放电设备生成一激励信号,以使待测电池生成激励响应信号;
    根据电池激励响应信号得到所述测试数据。
  5. 根据权利要求4所述的电池容量预测方法,其特征在于,基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数,包括:
    基于特征提取模型截取脉冲充放电过程中的测试数据,对测试数据进行分节处理得到多个分节电压值,并提取电池脉冲测试完成静置后与所述多个分节电压值对应的多个静置电压值,基于所述分节电压值和静置电压值计算充放电欧姆内阻和充放电极化内阻,得到脉冲充放电测试数据;
    基于特征提取模型截取短时间充电过程中的测试数据,确定充电时间、平台电压、开路电压、电流值和静置后的启止电压值,并根据充电时间和电流值计算充电电量,根据启止电压值计算回弹电压,得到短时间充电测试数据;
    根据脉冲充放电测试数据和短时间充电测试数据,得到所述待测电池第一特征参数。
  6. 根据权利要求5所述的电池容量预测方法,其特征在于,第一特征参数包括:充放电欧姆内阻、充放电极化内阻、开路电压、充电容量、平台电压、充电时间及回弹电压。
  7. 根据权利要求1所述的电池容量预测方法,其特征在于,基于特征降噪模型对所述第一特征参数进行降噪处理,得到第二特征参数,包括:
    基于所述降噪模型以所述待测电池的第一特征参数为输入得到特征参数的相关性系数,根据相关性系数对所述第一特征参数进行特征缩减,得到待测电池的第二特征参数。
  8. 根据权利要求1所述的电池容量预测方法,其特征在于,以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库,包括:
    基于神经网络、或支持向量机、或深度学习训练得到所述容量预测模型;
    基于容量预测模型,以所述待测电池的第二特征参数为输入,得到待测电 池的容量预估值;
    将待测电池的基本信息、第二特征参数和容量预估值存入样本数据库。
  9. 一种电池容量预测装置,其特征在于,包括:
    信息获取单元,用于获取待测电池的基本信息;
    模型确定单元,用于根据所述基本信息确定所述待测电池的特征提取模型、特征降噪模型以及容量预测模型;
    电池测试单元,用于获取待测电池进行充放电测试的测试数据;
    数据截取单元,用于基于特征提取模型对所述测试数据进行有效截取,得到待测电池的第一特征参数;
    参数降噪单元,用于基于特征降噪模型对所述第一特征参数进行降噪处理,得到第二特征参数;
    容量预估单元,用于以第二特征参数作为容量预测模型的输入,输出得到待测电池的容量预估值,并存入样本数据库。
  10. 一种电子设备,其特征在于,包括存储器和处理器,其中,
    所述存储器,用于存储程序;
    所述处理器,与所述存储器耦合,用于执行所述存储器中存储的所述程序,以实现上述权利要求1至8中任意一项所述的电池容量预测方法中的步骤。
PCT/CN2022/143343 2022-11-02 2022-12-29 一种电池容量预测方法、装置及电子设备 WO2024093005A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211361376.6 2022-11-02
CN202211361376.6A CN115656834A (zh) 2022-11-02 2022-11-02 一种电池容量预测方法、装置及电子设备

Publications (1)

Publication Number Publication Date
WO2024093005A1 true WO2024093005A1 (zh) 2024-05-10

Family

ID=84995025

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/143343 WO2024093005A1 (zh) 2022-11-02 2022-12-29 一种电池容量预测方法、装置及电子设备

Country Status (2)

Country Link
CN (1) CN115656834A (zh)
WO (1) WO2024093005A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116794533B (zh) * 2023-08-21 2023-12-29 深圳海辰储能科技有限公司 一种电芯容量分档方法和相关产品

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108983103A (zh) * 2018-06-29 2018-12-11 上海科列新能源技术有限公司 一种动力电池的数据处理方法和装置
CN112098848A (zh) * 2020-09-14 2020-12-18 北京大学深圳研究生院 一种电池放电容量预测方法、系统及可读存储介质
CN112213643A (zh) * 2020-09-30 2021-01-12 蜂巢能源科技有限公司 电池初始容量、电池健康状态的预测方法、系统及设备
CN113049974A (zh) * 2021-02-02 2021-06-29 欣旺达电动汽车电池有限公司 电池容量预测方法、装置、设备及存储介质
CN113406496A (zh) * 2021-05-26 2021-09-17 广州市香港科大霍英东研究院 基于模型迁移的电池容量预测方法、系统、装置及介质
CN113687250A (zh) * 2021-08-18 2021-11-23 蜂巢能源科技有限公司 电芯容量预测方法、装置、电子设备及介质
CN113848480A (zh) * 2021-09-23 2021-12-28 广东恒翼能科技有限公司 锂电池分容工艺的放电总容量预测方法
WO2022035153A1 (ko) * 2020-08-13 2022-02-17 주식회사 엘지에너지솔루션 배터리 용량 추정 장치 및 방법
CN114660468A (zh) * 2022-03-29 2022-06-24 武汉动力电池再生技术有限公司 一种退役电池荷电状态预测方法、装置及电子设备
CN115166532A (zh) * 2022-07-25 2022-10-11 武汉动力电池再生技术有限公司 镍氢电池容量预测方法、装置、电子设备及存储介质

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108983103A (zh) * 2018-06-29 2018-12-11 上海科列新能源技术有限公司 一种动力电池的数据处理方法和装置
WO2022035153A1 (ko) * 2020-08-13 2022-02-17 주식회사 엘지에너지솔루션 배터리 용량 추정 장치 및 방법
CN112098848A (zh) * 2020-09-14 2020-12-18 北京大学深圳研究生院 一种电池放电容量预测方法、系统及可读存储介质
CN112213643A (zh) * 2020-09-30 2021-01-12 蜂巢能源科技有限公司 电池初始容量、电池健康状态的预测方法、系统及设备
CN113049974A (zh) * 2021-02-02 2021-06-29 欣旺达电动汽车电池有限公司 电池容量预测方法、装置、设备及存储介质
CN113406496A (zh) * 2021-05-26 2021-09-17 广州市香港科大霍英东研究院 基于模型迁移的电池容量预测方法、系统、装置及介质
CN113687250A (zh) * 2021-08-18 2021-11-23 蜂巢能源科技有限公司 电芯容量预测方法、装置、电子设备及介质
CN113848480A (zh) * 2021-09-23 2021-12-28 广东恒翼能科技有限公司 锂电池分容工艺的放电总容量预测方法
CN114660468A (zh) * 2022-03-29 2022-06-24 武汉动力电池再生技术有限公司 一种退役电池荷电状态预测方法、装置及电子设备
CN115166532A (zh) * 2022-07-25 2022-10-11 武汉动力电池再生技术有限公司 镍氢电池容量预测方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN115656834A (zh) 2023-01-31

Similar Documents

Publication Publication Date Title
CN108228325B (zh) 应用管理方法和装置、电子设备、计算机存储介质
CN111679199B (zh) 一种锂离子电池soc估计方法及装置
CN108985060A (zh) 一种大规模安卓恶意软件自动化检测系统及方法
Feng et al. State of health estimation of large-cycle lithium-ion batteries based on error compensation of autoregressive model
WO2024093005A1 (zh) 一种电池容量预测方法、装置及电子设备
CN109633448B (zh) 识别电池健康状态的方法、装置和终端设备
CN109800446B (zh) 一种锂离子电池放电过程电压不一致性估计方法及装置
CN114881343B (zh) 基于特征选择的电力系统短期负荷预测方法及装置
CN107992210A (zh) 输入法词汇推荐方法、智能终端及具有存储功能的装置
CN116699446A (zh) 退役电池快速分选方法、装置、设备及存储介质
CN112084459A (zh) 电池充放电循环寿命的预测方法、装置、电子终端、及存储介质
WO2024060539A1 (zh) 动力电池的年度碳排放量估算方法及装置
CN117289167A (zh) 基于多重神经网络的电池剩余寿命预测方法、装置及介质
CN110704614B (zh) 对应用中的用户群类型进行预测的信息处理方法及装置
CN115454466A (zh) 机器学习模型自动更新的方法、装置、设备和介质
CN113125965B (zh) 电池析锂检测方法、装置、设备及存储介质
CN109436980B (zh) 电梯部件的状态检测方法和系统
CN110177006B (zh) 基于接口预测模型的节点测试方法及装置
CN115496120A (zh) 物联网设备异常检测方法、装置、设备及存储介质
CN115291111B (zh) 电池静置时间预测模型的训练方法以及静置时间预测方法
CN117112445B (zh) 一种机器学习模型稳定性检测方法、装置、设备及介质
WO2023004730A1 (en) Creating a target model for detecting legitimacy of data collection
CN107360230A (zh) 一种应用下载方法、装置和计算机可读存储介质
CN110542851A (zh) 断路器操作机构故障诊断方法、装置、计算机和存储介质
CN117371506A (zh) 模型训练、模型测试方法、装置、电子设备及存储介质