技术领域Technical field
本发明涉及短路电流检测方法，具体涉及一种电池组内短路电流估计方法。The invention relates to a short-circuit current detection method, and in particular to a short-circuit current estimation method in a battery pack.
背景技术Background technique
锂离子电池由于其较高的能量密度、功率密度、较长的使用寿命与环境友好的特点被广泛应用于电动汽车、储能系统以及便携式移动设备等领域。但是内短路是锂电池目前面对的较为严重的一种故障类型，当短路发生时，严重可以导致电池在短时间内起火造成人员伤亡，因此对于电池安全状态的检测是十分必要且重要的，尤其是短路电流的检测。Lithium-ion batteries are widely used in electric vehicles, energy storage systems, portable mobile devices and other fields due to their high energy density, power density, long service life and environmental friendliness. However, internal short circuit is a more serious type of fault that lithium batteries currently face. When a short circuit occurs, it can cause the battery to catch fire in a short time and cause casualties. Therefore, it is very necessary and important to detect the safety status of the battery. Especially the detection of short circuit current.
现存方法主要是对短路进行定性判断或者添加装置来检测短路电流是否发生及其大小，在[1]中通过获取电池电芯的电压进而计算各个电芯的残差，通过设定阈值的方法判断电池是否发生短路，依赖于异常阈值的设定以及判定的标准，当阈值设置较低时，容易将正常电池检测为异常；当阈值高时，会将异常电池判定为正常。对判定标准有较高的精度要求。在[2]中设计了锂电池电芯短路测试仪，检查锂电池在某生产环节的正负极之间是否存在短路或微短路情况，使用额外的测定装置来检测电池短路情况。Existing methods mainly make qualitative judgments on short circuits or add devices to detect whether short-circuit current occurs and its magnitude. In[1] , the voltage of battery cells is obtained and the residual error of each cell is calculated, and the judgment is made by setting a threshold. Whether a short circuit occurs in a battery depends on the setting of the abnormal threshold and the judgment standard. When the threshold is set low, it is easy to detect a normal battery as abnormal; when the threshold is high, the abnormal battery is judged as normal. There are higher accuracy requirements for the judgment standards. In[2] , a lithium battery cell short circuit tester is designed to check whether there is a short circuit or micro short circuit between the positive and negative electrodes of the lithium battery in a certain production process, and an additional measuring device is used to detect the battery short circuit.
现有的短路电流检测方法基于特定的检测装置或进行额外的测定过程，如特定的逻辑电路结构，现存的多种检测方法存在成本和检测精度的问题，比较难以得到较高精度的电流值。Existing short-circuit current detection methods are based on specific detection devices or additional measurement processes, such as specific logic circuit structures. Various existing detection methods have problems with cost and detection accuracy, making it difficult to obtain higher-precision current values.
参考文献：references:
[1]朱广焱,张鹏博,施璐等.锂电池组内部短路异常诊断方法及系统[P].上海市：CN115621584A,2023-01-17.[1] Zhu Guangyan, Zhang Pengbo, Shi Lu, et al. Diagnosis method and system for internal short circuit abnormality of lithium battery pack [P]. Shanghai: CN115621584A, 2023-01-17.
[2]张庆祥,李大印.锂电池电芯短路测试仪[P].CN307762444S,2022-12-30。[2] Zhang Qingxiang, Li Dayin. Lithium battery cell short circuit tester [P]. CN307762444S, 2022-12-30.
发明内容Contents of the invention
本发明的目的在于提供一种电池组内短路电流估计方法，以基于检测的电压实现电池组内的电池短路时短路电流的检测计算。The object of the present invention is to provide a method for estimating short-circuit current in a battery pack, so as to detect and calculate the short-circuit current when the batteries in the battery pack are short-circuited based on the detected voltage.
为了实现上述目的，本发明提供一种电池组内短路电流估计方法，包括：In order to achieve the above objectives, the present invention provides a method for estimating short-circuit current in a battery pack, including:
S1：选取等效电路模型来建立正常的单电池的模型，得到单电池的模型的状态空间方程；以单电池的模型串联，且在其中一个单电池的模型中加入短路支路以得到短路电流电池模型，来搭建串联电池组的仿真模型；S1: Select the equivalent circuit model to establish a normal single-cell model, and obtain the state space equation of the single-cell model; connect the single-cell models in series, and add a short-circuit branch to one of the single-cell models to obtain the short-circuit current. Battery model to build a simulation model of a series battery pack;
S2：根据所述单电池的模型的状态空间方程和单个电池的采集数据，所述采集数据包括端电压和电流，进行参数辨识以得到单个电池的特性参数；S2: According to the state space equation of the single battery model and the collected data of a single battery, where the collected data includes terminal voltage and current, perform parameter identification to obtain the characteristic parameters of a single battery;
S3：提供正常的电池组成的实际的串联电池组；对于其中的每个电池，在单电池的模型下，根据单个电池的特性参数，以指定的电流作为输入，实时采集的端电压作为观测量，使用容积卡尔曼滤波方法估计该电池的SOC，作为其SOC初值；S3: Provides an actual series battery pack composed of normal batteries; for each battery, under the single battery model, according to the characteristic parameters of the single battery, the specified current is used as the input, and the terminal voltage collected in real time is used as the observation quantity , use the volumetric Kalman filter method to estimate the SOC of the battery as its initial SOC value;
S4：采集实际的串联电池组在指定的电流下，每个电池的实时端电压；对于每个电池，在短路电流电池模型中，根据电池的实时端电压和SOC初值，使用无迹卡尔曼滤波来估计短路电流，并根据估计的短路电流判断该电池是否短路。S4: Collect the real-time terminal voltage of each battery in the actual series battery pack at the specified current; for each battery, in the short-circuit current battery model, use unscented Kalman based on the real-time terminal voltage of the battery and the initial SOC value. Filter to estimate the short-circuit current, and determine whether the battery is short-circuited based on the estimated short-circuit current.
所述单电池是锂电池，且选取的等效电路模型是一阶RC模型。The single battery is a lithium battery, and the equivalent circuit model selected is a first-order RC model.
一阶RC模型的状态空间方程为：The state space equation of the first-order RC model is:
其中，τ是时间常数，τ＝Cp×Rp，Cp是极化电容、Rp是极化电阻，Up是极化电压，R0是欧姆内阻，UOC是开路电压，Ut是端电压，It是外部电流。Among them, τ is the time constant, τ = Cp × Rp , Cp is the polarization capacitance, Rp is the polarization resistance, Up is the polarization voltage, R0 is the ohmic internal resistance, UOC is the open circuit voltage, Ut is the terminal voltage, It is the external current.
在所述步骤S2中，单个电池的采集数据来自于城市测力计驾驶计划的公开数据集。In step S2, the collected data of a single battery comes from the public data set of the urban ergometer driving plan.
在所述步骤S2中，进行参数辨识所采用的辨识方法是遗忘因子递推最小二乘法。In step S2, the identification method used for parameter identification is the forgetting factor recursive least squares method.
在进行参数辨识之前，还包括所述单电池的模型的状态空间方程表示成离散化的状态空间方程，以实现基于回归模型Y＝aX+v的参数辨识。Before parameter identification, the state space equation of the single cell model is also expressed as a discretized state space equation to achieve parameter identification based on the regression model Y=aX+v.
在所述步骤S2中，单个电池的特性参数包括极化电容、极化电阻、欧姆内阻和开路电压。In step S2, the characteristic parameters of a single battery include polarization capacitance, polarization resistance, ohmic internal resistance and open circuit voltage.
所述步骤S3具体包括：The step S3 specifically includes:
S31：对于每个电池，以k时刻的外部电流It,k作为输入，k时刻的端电压Ut,k作为观测量，以k时刻的SOC值SOCk、k时刻的极化电压Up,k作为状态量，得到基于单电池的模型的状态方程与观测方程；S31: For each battery, the external current It,k at time k is used as the input, the terminal voltage Ut,k at time k is used as the observation quantity, the SOC value SOCk at time k and the polarization voltage Up at time k are used,k is used as the state quantity to obtain the state equation and observation equation of the single cell-based model;
S32：根据基于单电池的模型的状态方程与观测方程，使用容积卡尔曼滤波方法进行SOC估计；S32: Based on the state equation and observation equation of the single-cell-based model, use the volumetric Kalman filter method to estimate SOC;
其中，基于单电池的模型的状态方程与观测方程为：Among them, the state equation and observation equation of the single-cell-based model are:
式中，Δt表示系统的采样频率，UOC(SOCk)为基于k时刻SOC值计算得到的开路电压，R0,k、Rp,k、τk、Up,k是k时刻的一阶RC模型的欧姆内阻、极化电容、时间常数和极化电压，It,k是k时刻的外部电流。In the formula, Δt represents the sampling frequency of the system, UOC (SOCk ) is the open circuit voltage calculated based on the SOC value at time k, and R0,k , Rp,k , τk , and Up,k are the values at time k. The ohmic internal resistance, polarization capacitance, time constant and polarization voltage of the first-order RC model, It,k is the external current at time k.
所述步骤S4具体包括：以极化电压、开路电压、SOC以及短路电流作为状态量，以端电压作为输入，以电流作为观测量，建立短路电流电池模型的状态空间方程。The step S4 specifically includes: using polarization voltage, open-circuit voltage, SOC and short-circuit current as state quantities, terminal voltage as input, and current as an observation quantity, establishing a state space equation of the short-circuit current battery model.
所述短路电流电池模型的状态空间方程为：The state space equation of the short-circuit current battery model is:
式中：δ1为中间参数，下标k表示采样时刻，Δt为采样间隔，R0表示欧姆内阻，R1和C1表示极化内阻和极化电容，V和I分别表示测量的电池的端电压和电流，V1表示极化电压，VOC是开路电压，S表示SOC，If表示短路电流，Q0表示电池容量，dVOC/dS表示开路电压对于SOC的导数，Imf表示测量的电池外部电流，w和v分别表示过程噪声和测量噪声。In the formula: δ1 is the intermediate parameter, The subscript k represents the sampling time, Δt is the sampling interval, R0 represents the ohmic internal resistance, R1 and C1 represent the polarization internal resistance and polarization capacitance, V and I represent the measured terminal voltage and current of the battery respectively, V1 represents the polarization voltage, VOC is the open circuit voltage, S represents SOC, If represents the short circuit current, Q0 represents the battery capacity, dVOC /dS represents the derivative of the open circuit voltage with respect to SOC, Imf represents the measured battery external current, w and v represents process noise and measurement noise respectively.
本发明的电池组内短路电流估计方法通过电池的等效电路模型和电池组的仿真模型来确定短路电流与电压之间的关系，实时提取锂电池的电流电压数据，以电压作为输入，电流作为观测量，将短路电流的部分单独提取出来作为短路电流电池模型的状态空间方程的状态量一部分，来检测短路电流的值，当电池短路时，端电压会因此收到影响，进而通过建立的模型反映到短路电流上来，因此不需要特殊的测定装置来测量短路电流，仅仅是基于模型层面的改进；同时，由于短路电流是最终得到的状态量，因此能够给出短路电流的具体值而非单纯判断是否短路。The short-circuit current estimation method in the battery pack of the present invention determines the relationship between short-circuit current and voltage through the equivalent circuit model of the battery and the simulation model of the battery pack, extracts the current and voltage data of the lithium battery in real time, uses the voltage as input, and the current as Observation quantity, the short-circuit current part is extracted separately as part of the state quantity of the state space equation of the short-circuit current battery model to detect the value of the short-circuit current. When the battery is short-circuited, the terminal voltage will be affected by this, and then through the established model It is reflected on the short-circuit current, so there is no need for a special measuring device to measure the short-circuit current. It is just an improvement based on the model level. At the same time, since the short-circuit current is the final state quantity, the specific value of the short-circuit current can be given rather than simply Determine whether there is a short circuit.
此外，本发明选用遗忘因子最小二乘法进行参数辨识，再通过容积卡尔曼滤波的方法估计SOC，基于SOC的估计结果去计算短路电流，使得短路电流电池模型的状态空间方程所需的SOC为已知结果，便于进行短路电流估计。本发明采用的模型为最小二乘法和卡尔曼滤波，模型较简单，可以在短时间内给出短路电流的估算结果。In addition, the present invention uses the forgetting factor least squares method for parameter identification, and then estimates the SOC through the volumetric Kalman filtering method, and calculates the short-circuit current based on the estimation results of the SOC, so that the SOC required for the state space equation of the short-circuit current battery model is Knowing the result facilitates short-circuit current estimation. The models used in this invention are the least square method and Kalman filtering. The model is relatively simple and can provide the estimation results of the short-circuit current in a short time.
本发明主要用于锂电池运行过程中的短路检测，当电池发生内短路时，可以根据测量到的电流电压检测短路电流的大小，为锂电池的正常运行提供保障。The invention is mainly used for short-circuit detection during the operation of lithium batteries. When an internal short-circuit occurs in the battery, the size of the short-circuit current can be detected based on the measured current and voltage, thereby ensuring the normal operation of the lithium battery.
附图说明Description of the drawings
图1是本发明的一种电池组内短路电流估计方法的流程图；Figure 1 is a flow chart of a method for estimating short-circuit current in a battery pack according to the present invention;
图2是本发明的电池组内短路电流估计方法采用的正常的单电池模型的结构示意图；Figure 2 is a schematic structural diagram of a normal single cell model used in the short-circuit current estimation method in the battery pack of the present invention;
图3A-图3C是对于单个正常电池，UDDS的测试结果图；Figure 3A-Figure 3C are UDDS test results for a single normal battery;
图4是本发明的电池组内短路电流估计方法采用的短路电流电池模型的结构示意图；Figure 4 is a schematic structural diagram of the short-circuit current battery model used in the short-circuit current estimation method in the battery pack of the present invention;
图5A和图5B是短路电阻为10Ω时的短路电流计算值与SOC估计值图；Figure 5A and Figure 5B are graphs of the calculated short-circuit current value and estimated SOC value when the short-circuit resistance is 10Ω;
图6A和图6B是未短路的电池组的短路电流计算值与SOC估计值图。6A and 6B are graphs of calculated short-circuit current values and estimated SOC values of a battery pack that is not short-circuited.
具体实施方式Detailed ways
下面结合附图，给出本发明的较佳实施例，并予以详细描述。Below, preferred embodiments of the present invention are given and described in detail with reference to the accompanying drawings.
本发明提供一种电池内短路电流估计方法，其基于检测的电压来检测电池组内的短路电流，尤其是电动汽车中串联的锂离子电池组内的短路电流。本发明的电池内短路电流估计方法能够以采集的电池电压作为输入，来实时监控锂离子电池的输出电流和短路电流，当输出短路电流发生异常时可以及时反映并给出短路电流值的具体大小，以便根据短路严重情况做出不同的应对策略，在不添加额外测定装置的情况下完成短路电流的计算。The present invention provides a short-circuit current estimation method in a battery, which detects the short-circuit current in the battery pack based on the detected voltage, especially the short-circuit current in the lithium-ion battery pack connected in series in an electric vehicle. The battery short-circuit current estimation method of the present invention can use the collected battery voltage as input to monitor the output current and short-circuit current of the lithium-ion battery in real time. When the output short-circuit current is abnormal, it can promptly reflect and give the specific size of the short-circuit current value. , so that different response strategies can be made according to the seriousness of the short circuit, and the calculation of the short circuit current can be completed without adding additional measuring devices.
如图1所示，本发明的电池内短路电流估计方法包括以下步骤：As shown in Figure 1, the battery short-circuit current estimation method of the present invention includes the following steps:
步骤S1：选取等效电路模型来建立正常的单电池的模型，得到单电池的模型的状态空间方程；以单电池的模型串联，且在其中一个单电池的模型中加入短路支路以得到短路电流电池模型，来搭建串联电池组的仿真模型；Step S1: Select the equivalent circuit model to establish a normal single-cell model, and obtain the state space equation of the single-cell model; connect the single-cell models in series, and add a short-circuit branch to one of the single-cell models to obtain a short circuit. Current battery model to build a simulation model of a series battery pack;
等效电路模型是根据电池的特性用电路模拟的一种模型，包含了电阻电容等器件，是模拟电池特性比较常用的一种模型。The equivalent circuit model is a model that is simulated with a circuit based on the characteristics of the battery. It includes devices such as resistors and capacitors. It is a commonly used model for simulating battery characteristics.
在本实施例中，由于模拟的是锂电池特性，因此，选取的等效电路模型是一阶RC模型(Thevenin模型)。也就是说，仿真中的单个电池是用一阶RC实现的，因为其复杂度低且能较好的对应锂电池的输入输出。一阶RC模型如图2所示，图2中，Cp是极化电容、Rp是极化电阻，Up是极化电压，R0是欧姆内阻，UOC是开路电压OCV、Ut是端电压，It是外部电流。In this embodiment, since the characteristics of a lithium battery are simulated, the equivalent circuit model selected is a first-order RC model (Thevenin model). In other words, the single battery in the simulation is implemented with a first-order RC because its complexity is low and it can better correspond to the input and output of the lithium battery. The first-order RC model is shown in Figure 2. In Figure 2, Cp is the polarization capacitance, Rp is the polarization resistance, Up is the polarization voltage, R0 is the ohmic internal resistance, UOC is the open circuit voltage OCV, Ut is the terminal voltage, It is the external current.
一阶RC模型的状态空间方程如下：The state space equation of the first-order RC model is as follows:
其中，τ是时间常数，τ＝Cp×Rp，Cp是极化电容、Rp是极化电阻，Up是极化电压，R0是欧姆内阻，UOC是开路电压OCV、Ut是端电压，It是外部电流。Among them, τ is the time constant, τ = Cp × Rp , Cp is the polarization capacitance, Rp is the polarization resistance, Up is the polarization voltage, R0 is the ohmic internal resistance, UOC is the open circuit voltage OCV, Ut is the terminal voltage and It is the external current.
所得到的仿真模型以电流为输入参数，以端电压为输出参数。电流电压可以互为输入输出，一般的步骤如上文参数辨识或者SOC估计都是电流作为输入且电压作为输出的，但是本发明的最后的短路电流计算需要把电流当作输出，电压当作输入才能估算短路电流值。The obtained simulation model takes the current as the input parameter and the terminal voltage as the output parameter. Current and voltage can be input and output to each other. General steps such as parameter identification or SOC estimation above use current as input and voltage as output. However, the final short-circuit current calculation of the present invention requires current as output and voltage as input. Estimate the short circuit current value.
搭建串联电池组的仿真模型是在Simulink上实现的。The simulation model of building a series battery pack is implemented on Simulink.
在所述步骤S3中，以单电池的模型串联，且其中一个单电池的模型加入短路支路以得到短路电流电池模型，来搭建串联电池组的仿真模型，具体包括：把两个电池的单电池模型(即两个等效电路模型)串联形成回路，然后其中一个单电池模型不变，另一个单电池模型加入短路支路以得到短路电流电池模型，得到串联电池组的仿真模型。In the step S3, single battery models are connected in series, and a short-circuit branch is added to one of the single battery models to obtain a short-circuit current battery model to build a simulation model of the series battery pack, which specifically includes: connecting the single cells of the two batteries. The battery models (i.e., two equivalent circuit models) are connected in series to form a loop, then one of the single battery models remains unchanged, and the other single battery model adds a short-circuit branch to obtain a short-circuit current battery model, and a simulation model of the series battery pack is obtained.
在本实施例中，在短路支路中，短路电阻设置为10欧姆，因为端电压在3V左右，因此短路电流大概为0.3A。电池发生内短路时短路支路上的电阻为短路电阻，这里把它设置的比较高(10欧姆)是为了检测小电流情况下的估计准确度。In this embodiment, in the short-circuit branch, the short-circuit resistance is set to 10 ohms, because the terminal voltage is about 3V, so the short-circuit current is about 0.3A. When an internal short circuit occurs in the battery, the resistance on the short-circuit branch is the short-circuit resistance. It is set relatively high (10 ohms) here to detect the estimation accuracy under small current conditions.
步骤S2：根据所述单电池的模型的状态空间方程和单个电池的采集数据，所述采集数据包括端电压和电流，进行参数辨识以得到单个电池的特性参数，单个电池的特性参数包括极化电容、极化电阻、欧姆内阻和开路电压；Step S2: According to the state space equation of the single battery model and the collected data of the single battery, the collected data includes terminal voltage and current, perform parameter identification to obtain the characteristic parameters of the single battery. The characteristic parameters of the single battery include polarization. Capacitance, polarization resistance, ohmic internal resistance and open circuit voltage;
因为实际中我们只能得到电池的电流电压数据，我们需要先选取一个模型(这里是等效电路模型)，然后根据电流电压对单个电池的特性参数进行辨识(辨识就是得到一个具体的值)，才能实现输入一个电流得到电压或者输入一个电压得到电流，我们选取一阶RC模型后只是有了模型的结构，但是针对具体的电池我们需要得到模型中各个参数的值，即电池各个器件的参数值，这也就是参数辨识的过程。参数辨识完成了之后，才能实现输入一个电流值能得到一个电压值。Because in practice we can only get the current and voltage data of the battery, we need to first select a model (here is the equivalent circuit model), and then identify the characteristic parameters of a single battery based on the current and voltage (identification is to get a specific value), In order to input a current to get a voltage or input a voltage to get the current, we only have the structure of the model after selecting the first-order RC model, but for a specific battery, we need to get the values of each parameter in the model, that is, the parameter values of each device of the battery. , which is the process of parameter identification. After the parameter identification is completed, it is possible to input a current value and obtain a voltage value.
其中，单个电池的采集数据来自于UDDS(Urban Dynamometer Driving Schedule，城市测力计驾驶计划)的公开数据集。Among them, the collected data of a single battery comes from the public data set of UDDS (Urban Dynamometer Driving Schedule).
UDDS是基于美国城市道路循环的UDDS的测试数据，用来模拟锂电池实际运行过程中的数据情况。UDDS公开数据集是NASA官方提供的模拟真实道路情况下单个锂电池的电流电压情况，其提供了UDDS工况下的电流数据和端电压数据。图3A-图3C示出了对于单个正常电池，用UDDS公开数据集的电流数据作为输入时对应的端电压数据和估算得到的SOC，即UDDS的测试结果图。UDDS is based on the UDDS test data of urban road cycles in the United States, which is used to simulate the data situation during the actual operation of lithium batteries. The UDDS public data set is officially provided by NASA to simulate the current and voltage conditions of a single lithium battery under real road conditions. It provides current data and terminal voltage data under UDDS operating conditions. Figures 3A-3C show for a single normal battery, the corresponding terminal voltage data and the estimated SOC when using the current data of the UDDS public data set as input, that is, the test result diagram of UDDS.
这里的单个电池的采集数据中估算得到的SOC是用安时积分得到的，就是电流乘时间，是为了当作真实的SOC，用作比较。The SOC estimated in the collected data of a single battery here is obtained by ampere-hour integration, which is current multiplied by time. It is used as a real SOC for comparison.
进行参数辨识所采用的辨识方法是遗忘因子递推最小二乘法(FFRLS,ForgettingFactor Recursive Least Squares)。The identification method used for parameter identification is Forgetting Factor Recursive Least Squares (FFRLS).
LS算法是一种用于解决多元线性拟合问题的工具。经典LS算法需要获取并保存目标系统的所有量测值后才能进行求解，这一方法无疑会消耗巨大的计算资源，同时耗时较长，也不具有实时性。因此，递推最小二乘(RLS,Recursive Least Square)这一概念被提出，其采用类似动态规划的方式来实现递推形式的在线更新目标系统的参数。而RLS算法是一种具有无限记忆长度的算法，当在电池系统中使用时，越来越多的旧数据累积会导致参数递推结果无法很好反映电池特性的变化，因此该算法被进一步改进，遗忘因子递推最小二乘(FFRLS,Forgetting Factor Recursive Least Squares)被提出。这种方法使参数更新更倚重新数据，同时也有效解决了“数据饱和”这一问题。由此，实现了等效电路模型的参数辨识，辨识得到单个电池的特性参数。The LS algorithm is a tool used to solve multivariate linear fitting problems. The classic LS algorithm needs to obtain and save all measurement values of the target system before it can be solved. This method will undoubtedly consume huge computing resources, take a long time, and is not real-time. Therefore, the concept of Recursive Least Square (RLS) was proposed, which uses a method similar to dynamic programming to achieve recursive online update of the parameters of the target system. The RLS algorithm is an algorithm with infinite memory length. When used in a battery system, the accumulation of more and more old data will cause the parameter recursion results to be unable to reflect changes in battery characteristics well, so the algorithm was further improved. , Forgetting Factor Recursive Least Squares (FFRLS, Forgetting Factor Recursive Least Squares) was proposed. This method makes parameter updates more dependent on new data, and also effectively solves the problem of "data saturation". As a result, the parameter identification of the equivalent circuit model is realized, and the characteristic parameters of a single battery are identified.
一般来说，回归模型可以表示为：Y＝aX+v，Generally speaking, the regression model can be expressed as: Y=aX+v,
因此，在进行参数辨识之前，还包括所述单电池的模型的状态空间方程(即一阶RC模型的状态方程)表示成离散化的状态空间方程。由此，能够在之后用FFRLS方法进行参数辨识。Therefore, before parameter identification is performed, the state space equation of the model of the single cell (that is, the state equation of the first-order RC model) is expressed as a discretized state space equation. Therefore, parameter identification can be performed later using the FFRLS method.
对步骤S1中的状态空间方程离散化可得：Discretizing the state space equation in step S1 gives:
式中，Δt表示系统的采样频率，UOC(SOCk)为基于k时刻SOC(荷电状态)值计算得到的开路电压，R0,k、Rp,k、τk、Up,k是k时刻的一阶RC模型的欧姆内阻、极化电容、时间常数和极化电压，It,k是k时刻的外部电流，Ut,k是k时刻的端电压。In the formula, Δt represents the sampling frequency of the system, UOC (SOCk ) is the open circuit voltage calculated based on the SOC (state of charge) value at time k, R0,k , Rp,k , τk , Up,k are the ohmic internal resistance, polarization capacitance, time constant and polarization voltage of the first-order RC model at time k, It,k is the external current at time k, and Ut,k is the terminal voltage at time k.
因此，最终得到的离散化的状态空间方程为：Therefore, the final discretized state space equation is:
式中，Δt表示采样频率，Ut,k是k时刻的端电压，Ut,k-1是k-1时刻的端电压，It,k是k时刻的外部电流，It,k-1是k-1时刻的外部电流，UOC,k为k时刻的开路电压，R0、Rp和τ是欧姆内阻、极化电容和时间常数，对其进行辨识。In the formula, Δt represents the sampling frequency, Ut,k is the terminal voltage at time k, Ut,k-1 is the terminal voltage at time k-1, It,k is the external current at time k, It,k- 1 is the external current at time k-1, UOC, k is the open circuit voltage at time k, R0 , Rp and τ are the ohmic internal resistance, polarization capacitance and time constant, identify them.
至此，另Y＝Ut,k，a＝[UOC,k,a0,a1,a2]，X＝[1,It,k-1,-(Ut,k-Ut,k-1)/Δt,(It,k-It,k-1)/Δt]，则可以实现基于回归模型Y＝aX+v的参数辨识。其中，a0,a1,a2分别是R0+Rp，τ，τR0。So far, Y=Ut,k , a=[UOC,k ,a0,a1,a2], X=[1,It,k-1 ,-(Ut,k -Ut,k-1 )/Δt, (It,k -It,k-1 )/Δt], then parameter identification based on the regression model Y=aX+v can be achieved. Among them, a0, a1 and a2 are R0 +Rp , τ and τR0 respectively.
下面是FFRLS算法的递推式，我们上面将等效电路模型写成了Y＝aX的形式，接下来就可以使用下面的遗忘因子递推最小二乘公式进行参数辨识了。The following is the recursive formula of the FFRLS algorithm. We wrote the equivalent circuit model above in the form of Y=aX. Next, we can use the following forgetting factor recursive least squares formula for parameter identification.
其中，Kk表示增益矩阵；Pk表示协方差矩阵；θk表示待辨识的参数矩阵；表示输入向量；λ表示遗忘因子；I表示单位矩阵。结合上述内容，只需令Y＝Ut,k，θ＝A，/>即可实现电池参数的递推在线辨识。实现过程就是通过编程建立好上述各个参数之间的关系，然后每输入一个电流就得到一组参数。Among them, Kk represents the gain matrix; Pk represents the covariance matrix; θk represents the parameter matrix to be identified; represents the input vector; λ represents the forgetting factor; I represents the identity matrix. Combined with the above content, just let Y=Ut,k , θ=A,/> This enables recursive online identification of battery parameters. The implementation process is to establish the relationship between the above parameters through programming, and then get a set of parameters every time a current is input.
步骤S3：提供正常的电池组成的实际的串联电池组；对于其中的每个电池，在单电池的模型下，根据单个电池的特性参数，以指定的电流(即指定的UDDS电流)作为输入，实时采集的端电压作为观测量，使用容积卡尔曼滤波方法(CKF)估计该电池的SOC，作为SOC初值；Step S3: Provide an actual series battery pack composed of normal batteries; for each battery, under the single battery model, according to the characteristic parameters of the single battery, use the specified current (i.e. the specified UDDS current) as input, The terminal voltage collected in real time is used as an observation quantity, and the volumetric Kalman filtering method (CKF) is used to estimate the SOC of the battery as the initial SOC value;
在进行短路计算之前要先进行SOC的估计，因为在短路电流计算模型中，状态空间方程含有对SOC求导项，因此需要一个准确的SOC值作为前提，SOC作为短路电流模型的输入初值。Before performing short-circuit calculation, SOC must be estimated first, because in the short-circuit current calculation model, the state space equation contains a derivative term for SOC, so an accurate SOC value is required as a premise, and SOC is used as the input initial value of the short-circuit current model.
需要说明的是，在步骤S3中，根据UDDS电流数据输入到仿真模型的两个单电池模型中，可以获取两个正常的单电池的端电压，从而获得串联电池组还没有发生电路时每个单电池的端电压。这里的SOC估计中两个单电池的电流电压数据与UDDS工况下的电流电压是一致的，因为单电池模型是符合UDDS公开数据集中的电流电压对应关系的。步骤S3的SOC估计只是为了给短路电流计算提供一个准确的SOC初值，因此假设开始时刻是没有发生短路的。由于刚开始串联电池组还没有发生短路，在运行了大概十秒之后才接入短路通路，这时候已经获得了比较准确的SOC初值，该SOC初值是正常电池的SOC，可以给到短路电流模型中去计算短路电流，到这SOC初值估计的任务就结束了，后面就是用短路电流模型计算短路电流了。It should be noted that in step S3, according to the UDDS current data input into the two single cell models of the simulation model, the terminal voltages of the two normal single cells can be obtained, thereby obtaining the voltage of each of the series connected battery packs before the circuit occurs. The terminal voltage of a single cell. The current and voltage data of the two single cells in the SOC estimation here are consistent with the current and voltage under UDDS operating conditions, because the single cell model is consistent with the current and voltage correspondence in the UDDS public data set. The SOC estimation in step S3 is only to provide an accurate initial SOC value for short-circuit current calculation, so it is assumed that no short circuit occurs at the beginning. Since the series-connected battery pack has not yet experienced a short circuit at the beginning, it was connected to the short-circuit path after running for about ten seconds. At this time, a relatively accurate initial SOC value has been obtained. This initial SOC value is the SOC of a normal battery and can be used to detect short circuits. To calculate the short-circuit current in the current model, the task of estimating the initial SOC value is over. The next step is to use the short-circuit current model to calculate the short-circuit current.
使用容积卡尔曼滤波方法估计SOC实际上和后面短路电流计算模型比较像，都是先建立模型(一阶RC模型)，想要估计哪个参数就把该参数写到状态空间方程的状态量里，然后用卡尔曼滤波族的方法去估计，只不过短路电流计算模型多了一个短路支路，而且是电压作为输入电流作为输出的(一般都是电流作为输入电压作为输出，但是电压作为输入电流作为输出也可以，因为主要估计的量是状态空间方程里的状态量，根据输入输出去不断矫正这个状态量)，数学表达式有差别。Using the volumetric Kalman filter method to estimate SOC is actually more similar to the short-circuit current calculation model later. They both establish a model (first-order RC model) first. If you want to estimate which parameter you want to estimate, write the parameter into the state quantity of the state space equation. Then use the Kalman filter family method to estimate, except that the short-circuit current calculation model has an additional short-circuit branch, and the voltage is used as the input current as the output (generally, the current is used as the input voltage as the output, but the voltage is used as the input current as the Output is also possible, because the main estimated quantity is the state quantity in the state space equation, and this state quantity is continuously corrected according to the input and output), and the mathematical expressions are different.
容积卡尔曼滤波方法的核心是根据球面径向容积准则，使用一组容积点来逼近非线性系统的状态均值和协方差。理论上来说，容积卡尔曼滤波方法(CKF)是最为接近贝叶斯滤波的近似算法，是解决非线性系统状态估算问题的有力工具。The core of the volume Kalman filter method is to use a set of volume points to approximate the state mean and covariance of the nonlinear system according to the spherical radial volume criterion. Theoretically speaking, the volumetric Kalman filtering method (CKF) is the closest approximation algorithm to Bayesian filtering and is a powerful tool for solving nonlinear system state estimation problems.
所述步骤S3具体包括：The step S3 specifically includes:
步骤S31：对于每个电池，以k时刻的外部电流It,k作为输入，k时刻的端电压Ut,k作为观测量，以k时刻的SOC值SOCk，k时刻的极化电压Up,k作为状态量(即令xk＝[SOCk Up,k]T)，得到基于单电池的模型的状态方程与观测方程；Step S31: For each battery, take the external current It,k at time k as the input, the terminal voltage Ut,k at time k as the observation quantity, the SOC value SOCk at time k, and the polarization voltage U at time k.p,k are used as state quantities (that is, let xk = [SOCk Up,k ]T ), and the state equation and observation equation of the single-cell-based model are obtained;
基于单电池的模型的状态方程与观测方程为：The state equation and observation equation of the single-cell-based model are:
式中，Δt表示系统的采样频率，UOC(SOCk)为基于k时刻SOC(荷电状态)值计算得到的开路电压，R0,k、Rp,k、τk、Up,k是k时刻的一阶RC模型的欧姆内阻、极化电容、时间常数和极化电压，It,k是k时刻的外部电流，uk表示一般的输入，这里指的是电流，Qcap是电池容量，η是库伦效率，通常是1。In the formula, Δt represents the sampling frequency of the system, UOC (SOCk ) is the open circuit voltage calculated based on the SOC (state of charge) value at time k, R0,k , Rp,k , τk , Up,k is the ohmic internal resistance, polarization capacitance, time constant and polarization voltage of the first-order RC model at time k, It,k is the external current at time k, uk represents the general input, here refers to the current, Qcap is the battery capacity, eta is the Coulomb efficiency, usually 1.
步骤S32：根据基于单电池的模型的状态方程与观测方程，使用容积卡尔曼滤波方法进行SOC估计。Step S32: Use the volumetric Kalman filter method to estimate SOC according to the state equation and observation equation of the single-cell-based model.
容积卡尔曼滤波方法的公式如下：The formula of the volumetric Kalman filter method is as follows:
实际上我们只要把模型数学表达写成上述状态空间方程的形式，把所需要估计的状态用状态量xk表示，就能用卡尔曼滤波族的方法去估算状态。步骤S31得到的基于单电池的模型的状态方程与观测方程就是上述形式的。In fact, as long as we write the mathematical expression of the model in the form of the above-mentioned state space equation, and express the state to be estimated by the state quantity xk , we can use the Kalman filter family method to estimate the state. The state equation and observation equation of the single-cell-based model obtained in step S31 are in the above form.
步骤S4：采集实际的串联电池组在指定的电流下，每个电池的实时端电压；对于每个电池，在短路电流电池模型中，根据电池的实时端电压和SOC初值，使用无迹卡尔曼滤波来估计短路电流，并根据估计的短路电流判断该电池是否短路。Step S4: Collect the real-time terminal voltage of each battery in the actual series-connected battery pack at the specified current; for each battery, in the short-circuit current battery model, use the unscented Karl method based on the real-time terminal voltage of the battery and the initial SOC value. Mann filtering is used to estimate the short-circuit current, and based on the estimated short-circuit current, it is determined whether the battery is short-circuited.
由此，本发明可以通过短路电流电池模型来实时判断各个电池是否发生短路。Therefore, the present invention can determine whether each battery is short-circuited in real time through the short-circuit current battery model.
电池的实时端电压是可以采集得到的。具体来说，只需要在每个电池两端设置一个电压测量器件就可以直接采集到该电池的实时端电压。The real-time terminal voltage of the battery can be collected. Specifically, you only need to set up a voltage measurement device at both ends of each battery to directly collect the real-time terminal voltage of the battery.
对于短路电流估算过程，实际上就是估计的方法改变了(从容积卡尔曼变成了无迹卡尔曼)，状态x里面的变量改变了，然后输入输出置换就可以了。For the short-circuit current estimation process, the estimation method is actually changed (from volumetric Kalman to unscented Kalman), the variables in state x are changed, and then the input and output are replaced.
其中，正常的单电池的模型如图2所示，短路电流电池模型如图4所示。在图4中，C1是极化电容、R1是极化电阻，VOC是开路电压，Imf是外部电流，If是短路电流，V是采集的电池电压，I是采集的电池电流，I等于Imf-If。Among them, the normal single battery model is shown in Figure 2, and the short-circuit current battery model is shown in Figure 4. In Figure 4, C1 is the polarization capacitor, R1 is the polarization resistor, VOC is the open circuit voltage, Imf is the external current,If is the short circuit current, V is the collected battery voltage, and I is the collected battery current. , I equals Imf -If .
所述步骤S4具体包括：以极化电压、开路电压、SOC以及短路电流作为状态量，以端电压作为输入，以电流作为观测量，建立短路电流电池模型的状态空间方程。The step S4 specifically includes: using polarization voltage, open-circuit voltage, SOC and short-circuit current as state quantities, terminal voltage as input, and current as an observation quantity, establishing a state space equation of the short-circuit current battery model.
假设短路电流If在相邻时间间隔内保持不变，则短路电流电池模型的状态空间方程为：Assuming that the short-circuit current If remains constant in adjacent time intervals, the state space equation of the short-circuit current battery model is:
式中：δ1为中间参数，In the formula: δ1 is the intermediate parameter,
下标k表示采样时刻(即第几步采样)，Δt为采样间隔，R0表示欧姆内阻，R1和C1表示极化内阻和极化电容，因为多了一个通路(短路电流)，相当于单电池模型变为了短路电流电池模型，所以用R1和C1代替Cp、Rp，实际上他们表示的部件是一样的。V和I分别表示电池的端电压和内部电流。V1表示极化电压，VOC是开路电压，S表示SOC，If表示短路电流，Q0表示电池容量。dVOC/dS表示开路电压对于SOC的导数，Imf表示测量的电池外部电流，即实际测量的电流，w和v分别表示过程噪声和测量噪声。The subscript k represents the sampling time (that is, which step of sampling), Δt is the sampling interval, R0 represents the ohmic internal resistance, R1 and C1 represent the polarization internal resistance and polarization capacitance, because there is an extra path (short circuit current) , which is equivalent to the single-cell model becoming a short-circuit current battery model, so R1 and C1 are used to replace Cp and Rp. In fact, they represent the same components. V and I represent the terminal voltage and internal current of the battery respectively. V1 represents the polarization voltage, VOC is the open circuit voltage, S represents SOC,If represents the short circuit current, and Q0 represents the battery capacity. dVOC /dS represents the derivative of the open circuit voltage with respect to SOC, Imf represents the measured battery external current, that is, the actual measured current, w and v represent process noise and measurement noise respectively.
需要说明的是，Imf和It表示的都是能用电流传感器测量到的电池外部电流，因为短路电流电池模型和普通的一阶RC模型在模型结构上发生了变化，所以用了不同字母表示。It should be noted that Imf and It both represent the external current of the battery that can be measured with a current sensor. Because the short-circuit current battery model and the ordinary first-order RC model have changed in the model structure, different letters are used. express.
因为短路电流电池模型的状态空间方程中含有SOC，因此一个准确的SOC初值是进行短路电流计算的关键。本发明通过上文的步骤S3先进行了SOC的估计，因此SOC是已知的。Because the state space equation of the short-circuit current battery model contains SOC, an accurate initial value of SOC is the key to calculating the short-circuit current. The present invention first estimates the SOC through the above step S3, so the SOC is known.
dVOC/dS表示开路电压对于SOC的导数，dVOC/dS这个参数是根据数据模拟估计得到的，电流Imf和电压V是实际采集到的，V就是电池的实时端电压。Q0表示电池容量，是已知的。R0、R1和C1是上文的步骤中通过参数辨识得到的(等于R0、Cp、Rp)。其余部分是算法迭代过程中自己估计的，我们只需要给一个初值。dVOC /dS represents the derivative of open circuit voltage with respect to SOC. The parameter dVOC /dS is estimated based on data simulation. The current Imf and voltage V are actually collected, and V is the real-time terminal voltage of the battery. Q0 represents the battery capacity, which is known. R0 , R1 and C1 are obtained through parameter identification in the above steps (equal to R0, Cp, Rp). The rest is estimated by ourselves during the iterative process of the algorithm, and we only need to give an initial value.
当电池没发生短路时，状态中的If为0。考虑到估计的短路电流会有波动，短路电流在几十毫安左右属于正常范围，依旧视为0，即短路电流的非零判断阈值为100毫安)。当电池短路时候，根据上述短路电流电池模型的状态空间方程可以得到相应的短路电流的值。由此，最终能够得到短路电流If。When there is no short circuit in the battery, If in the state is0 . Considering that the estimated short-circuit current will fluctuate, the short-circuit current is within the normal range of about tens of milliamperes and is still regarded as 0, that is, the non-zero judgment threshold of the short-circuit current is 100 milliamperes). When the battery is short-circuited, the corresponding short-circuit current value can be obtained according to the state space equation of the above short-circuit current battery model. Thus, the short-circuit currentIf can finally be obtained.
实验结果：Experimental results:
经过实验，本发明的效果和设计的预期一致。Through experiments, the effect of the present invention is consistent with the design expectations.
图5A、图5B给出短路电阻为10Ω时短路电流计算值与SOC估计值(端电压约3V)，图6A、图6B给出了未短路的电池组的短路电流计算值与SOC估计值，用于与之对比。图5B、图6B中一开始存在波动是估计的SOC，稳定变化是UDDS测试数据中的真实SOC。Figure 5A and Figure 5B show the calculated short-circuit current value and estimated SOC value when the short-circuit resistance is 10Ω (terminal voltage is about 3V). Figure 6A and Figure 6B show the calculated short-circuit current value and estimated SOC value of the battery pack that is not short-circuited. for comparison. In Figure 5B and Figure 6B, the initial fluctuation is the estimated SOC, and the stable change is the real SOC in the UDDS test data.
如图6A所示，未短路电池一开始的短路电流计算值也有一段是不为0这里属于刚开始估算电流算法处于收敛过程，在收敛后，短路电流计算值在0附近波动。因此，本发明可以有效地检测短路的发生，即使是零点几安的小电流。As shown in Figure 6A, the calculated short-circuit current value of the non-short-circuited battery at the beginning is not 0 for a period. This is because the current estimation algorithm is in the convergence process at the beginning. After convergence, the calculated short-circuit current value fluctuates around 0. Therefore, the present invention can effectively detect the occurrence of short circuit, even a small current of a few tenths of an amp.
以上所述的，仅为本发明的较佳实施例，并非用以限定本发明的范围，本发明的上述实施例还可以做出各种变化。凡是依据本发明申请的权利要求书及说明书内容所作的简单、等效变化与修饰，皆落入本发明专利的权利要求保护范围。本发明未详尽描述的均为常规技术内容。The above are only preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various changes can be made to the above-mentioned embodiments of the present invention. All simple and equivalent changes and modifications made based on the claims and description of the present invention fall within the scope of protection of the claims of the patent of the present invention. What is not described in detail in the present invention is conventional technical content.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291610.7ACN117572235A (en) | 2023-10-08 | 2023-10-08 | Method for estimating short-circuit current in battery pack |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291610.7ACN117572235A (en) | 2023-10-08 | 2023-10-08 | Method for estimating short-circuit current in battery pack |
| Publication Number | Publication Date |
|---|---|
| CN117572235Atrue CN117572235A (en) | 2024-02-20 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311291610.7APendingCN117572235A (en) | 2023-10-08 | 2023-10-08 | Method for estimating short-circuit current in battery pack |
| Country | Link |
|---|---|
| CN (1) | CN117572235A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119627271A (en)* | 2024-12-04 | 2025-03-14 | 苏州星德胜智能电气有限公司 | Battery pack normal energy saving method and system based on battery pack loss detection |
| CN119916224A (en)* | 2025-04-02 | 2025-05-02 | 福建时代星云科技有限公司 | A method and terminal for identifying and predicting short circuit of battery circuit |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119627271A (en)* | 2024-12-04 | 2025-03-14 | 苏州星德胜智能电气有限公司 | Battery pack normal energy saving method and system based on battery pack loss detection |
| CN119916224A (en)* | 2025-04-02 | 2025-05-02 | 福建时代星云科技有限公司 | A method and terminal for identifying and predicting short circuit of battery circuit |
| Publication | Publication Date | Title |
|---|---|---|
| Takyi‐Aninakwa et al. | A strong tracking adaptive fading‐extended Kalman filter for the state of charge estimation of lithium‐ion batteries | |
| CN113156321B (en) | A method for estimating the state of charge SOC of lithium-ion batteries | |
| CN110058160B (en) | Lithium battery state of health prediction method based on square root extended Kalman filter | |
| CN103323781B (en) | Power battery pack on-line parameter detection system and SOC method of estimation | |
| CN117572235A (en) | Method for estimating short-circuit current in battery pack | |
| CN111965547A (en) | Battery system sensor fault diagnosis method based on parameter identification method | |
| CN111190109B (en) | A method for estimating state of charge of lithium battery without current sensor | |
| CN113687251B (en) | Double-model-based lithium ion battery pack voltage abnormality fault diagnosis method | |
| CN105116344B (en) | Based on binary-coded battery open circuit voltage evaluation method | |
| CN111929581A (en) | Method for predicting internal and external temperatures of power lithium battery | |
| CN108828448B (en) | Battery state of charge on-line estimation method based on charging voltage curve fusion Kalman filtering | |
| Wang et al. | A hybrid model for state of charge estimation of lithium-ion batteries utilizing improved adaptive extended Kalman filter and long short-term memory neural network | |
| CN111965560A (en) | Battery health state estimation method oriented to general discharge working condition | |
| CN114740385A (en) | Self-adaptive lithium ion battery state of charge estimation method | |
| CN112946481A (en) | Based on federation H∞Filtering sliding-mode observer lithium ion battery SOC estimation method and battery management system | |
| CN107783057B (en) | Power battery SOC estimation method and estimation system | |
| CN111142025A (en) | Battery SOC estimation method and device, storage medium and electric vehicle | |
| CN110673037B (en) | Battery SOC estimation method and system based on improved simulated annealing algorithm | |
| CN117607704A (en) | Lithium ion battery pack micro-short circuit fault diagnosis method considering inconsistency | |
| Yang et al. | SOC estimation of lithium-ion battery based on new adaptive fading extended Kalman filter | |
| CN114355217A (en) | Leakage battery diagnosis method based on second-order equivalent circuit model | |
| Bage et al. | Enhanced moving-step unscented transformed-dual extended Kalman filter for accurate SOC estimation of lithium-ion batteries considering temperature uncertainties | |
| CN113985292B (en) | Lithium ion power battery SOC double-filter estimation method based on improved coupling mode | |
| Zhu et al. | Online state of charge EKF estimation for LiFePO 4 battery management systems | |
| CN112213642B (en) | Method for detecting and positioning disconnection of monomer in parallel battery pack |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |