技术领域Technical Field
本发明涉及新能源汽车的动力电池安全领域，具体涉及一种基于运行大数据的电动汽车动力电池实时安全预警方法。The present invention relates to the field of power battery safety of new energy vehicles, and in particular to a real-time safety early warning method for power batteries of electric vehicles based on operation big data.
背景技术Background Art
电动汽车因其高效的性能以及对解决温室气体排放和全球变暖等环境问题的贡献得到汽车业界的广泛认可。而动力电池是电动汽车主要的储能装置，因此，电池的安全可靠运行非常重要，与其他储能装置一样，电池即用即老化，主要表现在容量的衰减和功率衰退(内阻增加)。因此，电池的可靠运行建立在精准的电池参数、状态估计之上，而电池管理系统(Battery Management System,BMS)能够对电池状态实时监控。另一方面，随着大数据时代的到来，电动汽车、充电桩、云端大数据等车-边-端多场耦合监测电动汽车的安全性得到广泛的关注，如何利用大数据建立动力电池的安全模型是亟待解决的行业难题。Electric vehicles have been widely recognized by the automotive industry for their high efficiency and contribution to solving environmental problems such as greenhouse gas emissions and global warming. Power batteries are the main energy storage devices for electric vehicles. Therefore, the safe and reliable operation of batteries is very important. Like other energy storage devices, batteries age as soon as they are used, which is mainly manifested in the attenuation of capacity and power decay (increased internal resistance). Therefore, the reliable operation of batteries is based on accurate estimation of battery parameters and status, and the battery management system (BMS) can monitor the battery status in real time. On the other hand, with the advent of the big data era, the safety of electric vehicles, charging piles, cloud big data and other vehicle-edge-end multi-field coupling monitoring has received widespread attention. How to use big data to establish a safety model for power batteries is an industry problem that needs to be solved urgently.
当前对于动力电池的安全性研究大多建立在实验室的特定条件下，通过特定工况下精准的传感器的采集信息作为电池安全性的表征参数，建立经验模型、解析模型或神经网络模型，进而得到良好的状态量如直流内阻、荷电状态(Stateof Charge，SOC)、健康状态(StateofHealth，SOH)、峰值功率状态(StateofPower，SOP)。但实际车辆行驶数据与云端大数据平台监控的数据与实验室数据存在较大的差异，其具体特征表现为：1.运行大数据的采样频率往往是0.1Hz甚至更低，并且采样的频率不是一成不变的，这将导致相邻采样点的电池状态不连续甚至差异很大，而一些经典的状态辨识方法如递推最小二乘法进行参数辨识需要电池在单位采样间隔内消耗或吸收的电量对其SOC的影响近似为零、电池在单位采样间隔内温度不变等苛刻条件，因此这类依靠状态量缓慢变化的假设的方法无法运用到实车运行大数据当中。2.实验室可通过加速寿命试验在较大倍率下充放电快速完成动力电池的全寿命周期，而大数据平台监控的实车数据往往运行数十年才达到动力电池的寿命极限，将导致大数据平台监控下实车运行工况难以通过实验手段复现。3.由于实际运行工况、环境的复杂性，实验室条件下很难模拟出实际运行工况，并且，在多种不确定性条件的影响下，采集的数据会出现质量下降等一系列问题，在建模前需要数据清洗等复杂处理。At present, most of the safety research on power batteries is based on specific laboratory conditions. Through the accurate sensor information collected under specific working conditions as the characterization parameter of battery safety, empirical models, analytical models or neural network models are established to obtain good state quantities such as DC internal resistance, state of charge (SOC), state of health (SOH), and peak power state (SOP). However, there are large differences between the actual vehicle driving data and the data monitored by the cloud big data platform and the laboratory data. The specific characteristics are as follows: 1. The sampling frequency of the operation big data is often 0.1Hz or even lower, and the sampling frequency is not constant, which will cause the battery state of adjacent sampling points to be discontinuous or even very different. Some classic state identification methods such as recursive least squares method for parameter identification require that the influence of the power consumed or absorbed by the battery in the unit sampling interval on its SOC is approximately zero, and the battery temperature is constant in the unit sampling interval. Therefore, this kind of method that relies on the assumption that the state quantity changes slowly cannot be applied to the real vehicle operation big data. 2. The laboratory can quickly complete the full life cycle of the power battery by charging and discharging at a large rate through accelerated life tests, while the real vehicle data monitored by the big data platform often runs for decades before reaching the life limit of the power battery, which will make it difficult to reproduce the real vehicle operating conditions under the big data platform through experimental means. 3. Due to the complexity of the actual operating conditions and environment, it is difficult to simulate the actual operating conditions under laboratory conditions. In addition, under the influence of various uncertain conditions, the collected data will have a series of problems such as quality degradation, and complex processing such as data cleaning is required before modeling.
针对上述问题背景下，对于运行大数据的电动汽车，其开路电压(Open CircuitVoltage，OCV)随SOC的变化难以表征，内阻等关乎到电池安全的参数难以辨识。因此，亟需基于运行大数据的电动汽车动力电池实时安全预警方法。In view of the above problems, for electric vehicles running big data, it is difficult to characterize the change of open circuit voltage (OCV) with SOC, and parameters related to battery safety such as internal resistance are difficult to identify. Therefore, there is an urgent need for a real-time safety warning method for electric vehicle power batteries based on running big data.
发明内容Summary of the invention
为解决现有技术中的不足，本发明提供一种基于运行大数据的电动汽车动力电池实时安全预警方法，基于实车运行大数据，通过大量的运行数据有效得到OCV随SOC的变化曲线，进而进行充电和放电直流内阻的在线参数辨识，并建立有效的安全预警模型实现实时安全预警和检测可能出现故障的电池单体，实现动力电池系统的高安全性和高可靠性运行。In order to address the deficiencies in the prior art, the present invention provides a real-time safety warning method for electric vehicle power batteries based on operation big data. Based on the actual vehicle operation big data, a large amount of operation data is used to effectively obtain the OCV vs. SOC change curve, and then online parameter identification of charging and discharging DC internal resistance is performed, and an effective safety warning model is established to achieve real-time safety warning and detection of battery cells that may fail, thereby achieving high safety and high reliability operation of the power battery system.
本发明为实现上述目的，通过以下技术方案实现：一种基于运行大数据的电动汽车动力电池实时安全预警方法，包括以下步骤：To achieve the above-mentioned purpose, the present invention is implemented through the following technical solutions: a real-time safety early warning method for electric vehicle power batteries based on operation big data, comprising the following steps:
S1、数据读取，读取动力电池过往历史数据的总电流、总SOC、以及单体电压；S1, data reading, reading the total current, total SOC, and single cell voltage of the power battery's past historical data;
S2、数据清洗，针对缺失数据、重复值数据、错误数据进行清洗；S2, data cleaning, cleaning of missing data, duplicate value data, and erroneous data;
S3、数据分析，提取不同充电时刻的电压值，并记录对应时刻的SOC值，建立OCV-SOC曲线；S3, data analysis, extracting the voltage values at different charging times, and recording the SOC values at the corresponding times, and establishing the OCV-SOC curve;
S4、参数辨识，通过拟合得到的OCV-SOC曲线，对实时采集数据利用Rint模型进行参数辨识，得到充电段的直流内阻和放电段的直流内阻；S4, parameter identification, by fitting the obtained OCV-SOC curve, using the Rint model to perform parameter identification on the real-time collected data, and obtain the DC internal resistance of the charging section and the DC internal resistance of the discharging section;
S5、安全预警，对充电片段内阻在空间维度和时间维度上分别采用熵权重法和变异系数法进行安全预警，对放电片段内阻设置阈值预警。S5. Safety warning: Entropy weight method and coefficient of variation method are used to make safety warning for the internal resistance of the charging segment in the spatial dimension and time dimension respectively, and threshold warning is set for the internal resistance of the discharge segment.
进一步的，S2中，针对缺失值的数据量小于等于4个采样点，采用最近邻插值法对数据进行补全，对缺失的数据量大于4个采样点，直接剔除；针对重复值，仅保留一个有效值；针对内容有误的数据，当电压数据为零，则对该时刻的电流数据进行条件判断，若该时刻电流与上一采样点变化小于最大充放电流的3％，则电压采用上一采样点电压代替，若电流变化值大于最大充放电流的3％，则对该时刻电压直接剔除。Furthermore, in S2, for data with missing values less than or equal to 4 sampling points, the nearest neighbor interpolation method is used to complete the data, and for data with missing values greater than 4 sampling points, the data is directly eliminated; for repeated values, only one valid value is retained; for data with incorrect content, when the voltage data is zero, a conditional judgment is made on the current data at that moment. If the change in current at that moment compared with the previous sampling point is less than 3% of the maximum charge and discharge current, the voltage is replaced by the voltage of the previous sampling point. If the current change value is greater than 3% of the maximum charge and discharge current, the voltage at that moment is directly eliminated.
进一步的，S4中，建立哈希映射计算内阻，通过拟合得到的OCV-SOC曲线，可以得到每个SOC对应的OCV具体值，将SOC存储到哈希表的键向量keys里，将OCV存储到哈希表的值向量values里，得到实时采集的总电流、单体电压、总SOC，根据SOC的变化判断充电还是放电。Furthermore, in S4, a hash map is established to calculate the internal resistance. By fitting the obtained OCV-SOC curve, the specific OCV value corresponding to each SOC can be obtained. The SOC is stored in the key vector keys of the hash table, and the OCV is stored in the value vector values of the hash table. The total current, single cell voltage, and total SOC collected in real time are obtained, and charging or discharging is determined according to the change of SOC.
进一步的，对于充电过程中，在线辨识到每个单体的直流内阻，其计算式为：Furthermore, during the charging process, the DC internal resistance of each cell is identified online, and the calculation formula is:
OCV(k)＝Hashmap(SOC(k)) (I)OCV(k)＝Hashmap(SOC(k)) (I)
式(I)中OCV(k)是第k时刻的开路电压，SOC(k)是第k时刻的荷电状态，Hashmap为建立的SOC和OCV之间哈希映射，式(II)中，R(k)为第k时刻的直流内阻，Ut(k)为第k时刻的单体电池端电压，I(k)为第k时刻的单体电池电流值。In formula (I), OCV(k) is the open circuit voltage at the kth moment, SOC(k) is the state of charge at the kth moment, Hashmap is the hash mapping between SOC and OCV established, and in formula (II), R(k) is the DC internal resistance at the kth moment, Ut (k) is the single cell terminal voltage at the kth moment, and I(k) is the single cell current value at the kth moment.
进一步的，S4中，对于放电过程中，利用滤波算法结合最小二乘特征曲线间接得到放电段的直流内阻。Furthermore, in S4, during the discharge process, a filtering algorithm is used in combination with a least square characteristic curve to indirectly obtain the DC internal resistance of the discharge section.
进一步的，S5中充电片段在时间维度上预警方法为：Furthermore, the warning method of the charging segment in S5 in the time dimension is:
通过对电动汽车长期运行充电工况得知，其往往选择的充电策略为多阶恒流充电，但经统计发现，其在70％-90％SOC区间，往往采用大倍率充电的方式，并且该区间段内欧姆内阻的变化并不显著，而对于欧姆内阻变化不显著的区间段如若出现异常信息，则更容易检测到是否发生故障，因此选取此区间作为研究对象。对多个充电片段进行时间维度和空间不一致性维度的两步内阻一致性安全预警。Through the long-term operation of electric vehicles, we know that the charging strategy they often choose is multi-stage constant current charging. However, statistics show that in the 70%-90% SOC range, they often use a high rate charging method, and the change of ohmic internal resistance in this range is not significant. If abnormal information appears in the range where the ohmic internal resistance does not change significantly, it is easier to detect whether a fault has occurred. Therefore, this range is selected as the research object. Two-step internal resistance consistency safety warning is performed on multiple charging segments in the time dimension and spatial inconsistency dimension.
首先，选取所有单体在70％-90％SOC区间的内阻：First, select the internal resistance of all monomers in the 70%-90% SOC range:
其中，j表示单体电池个数，i表示不同SOC的第i个采样点；Wherein, j represents the number of single cells, and i represents the i-th sampling point of different SOCs;
然后，将矩阵R的每一行进行标准化处理，其计算式为：Then, each row of the matrix R is standardized, and the calculation formula is:
式中ri表示R矩阵的行向量，Rin表示第i行第n列的矩阵元素，经过处理后可得到无量纲的R_normal矩阵：Whereri represents the row vector of the R matrix, andRin represents the matrix element in the i-th row and n-th column. After processing, the dimensionless R_normal matrix can be obtained:
计算R_normal行向量的均值和标准差Mi和Si：Compute the mean and standard deviationMi andSi of the row vector of R_normal:
式中n表示电池单体的个数；Where n represents the number of battery cells;
最后，求得不同采样点的变异系数Vi：Finally, the coefficient of variationVi of different sampling points is obtained:
对于变异系数，结果在0.15以内认为是各个电池单体一致性良好，当某一时刻的变异系数大于0.15则表明该时刻出现电池单体的不一致。For the coefficient of variation, if the result is within 0.15, it is considered that the consistency of each battery cell is good. When the coefficient of variation at a certain moment is greater than 0.15, it indicates that the battery cell is inconsistent at that moment.
进一步的，S5中充电片段在空间维度上预警方法为：Furthermore, the early warning method of the charging segment in S5 in the spatial dimension is:
首先，选取n个电池单体，m个采样时刻，建立矩阵Xij(i＝1,2,…,n；j＝1,2,…,m)，然后对矩阵列向量进行归一化，计算第j个采样点下第i个电池单体占该时刻的比重pij：First, select n battery cells and m sampling moments, establish a matrixXij (i = 1, 2, ..., n; j = 1, 2, ..., m), then normalize the matrix column vector, and calculatethe proportion pij of the i-th battery cell at the j-th sampling point at that moment:
式中，Xij表示矩阵X中第i行第j列的值；WhereXij represents the value of the i-th row and j-th column in the matrix X;
然后，计算第j个时刻的熵值ej：Then, calculate the entropy value ej at the jth moment:
其中，根据熵值计算信息熵冗余度dj：in, Calculate the information entropy redundancy dj according to the entropy value:
dj＝1-ejdj =1-ej
对冗余度进行权值计算：Calculate the weight of redundancy:
利用每个单体计算所有j时刻下的总和可以得到各个电池单体的综合得分si：By calculating the sum of all j moments for each cell, we can get the comprehensive score si of each battery cell:
最后，将每个电池单体综合得分与所有单体得分的均值进行作差得到偏离度Δs，偏离度越大，反应该单体偏离大多数单体的程度越大，通过偏离的程度来判断该单体是否出现异常。Finally, the deviation Δs is obtained by subtracting the comprehensive score of each battery cell from the mean score of all cells. The larger the deviation, the greater the degree of deviation of the cell from the majority of cells. The degree of deviation is used to determine whether the cell is abnormal.
进一步的，S3中提取若干充电片段起始时刻、运行过程中电流为零超过30min的时刻、充电结束并充分静置后的时刻的电压值。Furthermore, in S3, voltage values are extracted at the start time of several charging segments, at the time when the current is zero for more than 30 minutes during operation, and at the time when charging is completed and the battery is fully left to stand.
对比现有技术，本发明的有益效果在于：Compared with the prior art, the beneficial effects of the present invention are:
1.现有的OCV-SOC曲线往往只是在实验室离线条件下进行OCV实验或者混合功率脉冲特性(Hybrid Pulse Power Characterization，HPPC)得到的，从实车中难以获得，而本发明通过车桩记录的实车数据可以不通过实验得到实车电池的OCV-SOC曲线，且充电的次数越多，还原出的OCV-SOC曲线越准确。1. The existing OCV-SOC curve is often obtained by conducting OCV experiments or hybrid power pulse characterization (HPPC) under laboratory offline conditions, which is difficult to obtain from a real vehicle. However, the present invention uses the real vehicle data recorded by the vehicle pile to obtain the OCV-SOC curve of the real vehicle battery without experiments, and the more times the battery is charged, the more accurate the restored OCV-SOC curve is.
2.本发明能够有效的在线辨识出充电过程和放电过程动力电池的直流内阻，计算复杂度小，准确性高；并且，基于内阻信息提出的时间空间双维度安全预警方法即能有效诊断出发生故障的具体时间，还能诊断出现故障的具体电池单体，有效的实现电池系统安全精确预警。2. The present invention can effectively identify the DC internal resistance of the power battery during the charging and discharging processes online, with low calculation complexity and high accuracy; and the time-space dual-dimensional safety warning method proposed based on the internal resistance information can not only effectively diagnose the specific time when the fault occurs, but also diagnose the specific battery cell where the fault occurs, effectively realizing the safety and accurate warning of the battery system.
附图说明BRIEF DESCRIPTION OF THE DRAWINGS
图1为本方案实施的流程图；Figure 1 is a flow chart for the implementation of this scheme;
图2为单体电压数据清洗前的结果；Figure 2 shows the result before the cell voltage data is cleaned;
图3为单体电压数据清洗后的结果；Figure 3 shows the result after cleaning the single cell voltage data;
图4为拟合的OCV-SOC曲线；Figure 4 shows the fitted OCV-SOC curve;
图5为辨识的充电内阻随SOC的分布；Figure 5 shows the distribution of the identified charging internal resistance with SOC;
图6为辨识的放电内阻随SOC的分布；Figure 6 shows the distribution of the identified discharge internal resistance with SOC;
图7为不同时刻的变异系数；Figure 7 shows the coefficient of variation at different times;
图8为不同单体的偏离度；Figure 8 shows the deviation of different monomers;
具体实施方式DETAILED DESCRIPTION
下面结合具体实施例，进一步阐述本发明。应理解，这些实施例仅用于说明本发明而不用于限制本发明的范围。此外应理解，在阅读了本发明讲授的内容之后，本领域技术人员可以对本发明作各种改动或修改，这些等价形式同样落于本申请所限定的范围。The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the application equally.
如图1所示，一种基于运行大数据的电动汽车动力电池实时安全预警方法，包括以下步骤：As shown in FIG1 , a real-time safety early warning method for electric vehicle power batteries based on operation big data includes the following steps:
S1、数据读取，读取动力电池过往历史数据的总电流、总SOC、以及单体电压；S1, data reading, reading the total current, total SOC, and single cell voltage of the power battery's past historical data;
S2、数据清洗，针对缺失数据、重复值数据、错误数据进行清洗；S2, data cleaning, cleaning of missing data, duplicate value data, and erroneous data;
S3、数据分析，提取不同充电时刻的电压值，并记录对应时刻的SOC值，建立OCV-SOC曲线；S3, data analysis, extracting the voltage values at different charging times, and recording the SOC values at the corresponding times, and establishing the OCV-SOC curve;
S4、参数辨识，通过拟合得到的OCV-SOC曲线，对实时采集数据利用Rint(电池等效电路模型)模型进行参数辨识，得到充电段的直流内阻和放电段的直流内阻；S4, parameter identification, by fitting the obtained OCV-SOC curve, using the Rint (battery equivalent circuit model) model to perform parameter identification on the real-time collected data, and obtain the DC internal resistance of the charging section and the DC internal resistance of the discharging section;
S5、安全预警，对充电片段内阻在空间维度和时间维度上分别采用熵权重法和变异系数法进行安全预警，对放电片段内阻设置阈值预警。S5. Safety warning: Entropy weight method and coefficient of variation method are used to make safety warning for the internal resistance of the charging segment in the spatial dimension and time dimension respectively, and threshold warning is set for the internal resistance of the discharge segment.
由于车桩大数据存在缺失值、重复值、内容有误等问题无法直接使用，需要开展数据清洗工作，清晰方法为：针对缺失值的数据量小于等于4个采样点，采用最近邻插值法对数据进行补全，对缺失的数据量大于4个采样点，直接剔除；针对重复值，仅保留一个有效值；针对内容有误的数据，当电压数据为零，则对该时刻的电流数据进行条件判断，若该时刻电流与上一采样点变化小于最大充放电流的3％，则电压采用上一采样点电压代替，若电流变化值大于最大充放电流的3％，则对该时刻电压直接剔除，清洗前后的数据如图2和图3所示，图2为清洗前，图3为清洗后。Since the big data of vehicle charging piles cannot be used directly due to missing values, repeated values, incorrect content and other problems, it is necessary to carry out data cleaning. The cleaning method is: for data with missing values less than or equal to 4 sampling points, the nearest neighbor interpolation method is used to complete the data. For data with missing values greater than 4 sampling points, the data is directly removed. For repeated values, only one valid value is retained. For data with incorrect content, when the voltage data is zero, a conditional judgment is made on the current data at that moment. If the change in current at that moment from the previous sampling point is less than 3% of the maximum charge and discharge current, the voltage is replaced by the voltage of the previous sampling point. If the current change value is greater than 3% of the maximum charge and discharge current, the voltage at that moment is directly removed. The data before and after cleaning are shown in Figures 2 and 3, Figure 2 is before cleaning, and Figure 3 is after cleaning.
进一步的，S4中，建立哈希映射计算内阻，通过拟合得到的OCV-SOC曲线，可以得到每个SOC对应的OCV具体值，将SOC存储到哈希表的键向量keys里，将OCV存储到哈希表的值向量values里，得到实时采集的总电流、单体电压、总SOC，根据SOC的变化判断充电还是放电。Furthermore, in S4, a hash map is established to calculate the internal resistance. By fitting the obtained OCV-SOC curve, the specific OCV value corresponding to each SOC can be obtained. The SOC is stored in the key vector keys of the hash table, and the OCV is stored in the value vector values of the hash table. The total current, single cell voltage, and total SOC collected in real time are obtained, and charging or discharging is determined according to the change of SOC.
进一步的，对于充电过程中，在线辨识到每个单体的直流内阻，其计算式为：Furthermore, during the charging process, the DC internal resistance of each cell is identified online, and the calculation formula is:
OCV(k)＝Hashmap(SOC(k))OCV(k)＝Hashmap(SOC(k))
式中OCV(k)是第k时刻的开路电压，SOC(k)是第k时刻的荷电状态，Hashmap为建立的SOC和OCV之间哈希映射，R(k)为第k时刻的直流内阻，Ut(k)为第k时刻的单体电池端电压。Where OCV(k) is the open circuit voltage at the kth moment, SOC(k) is the state of charge at the kth moment, Hashmap is the established hash mapping between SOC and OCV, R(k) is the DC internal resistance at the kth moment, and Ut (k) is the terminal voltage of the single cell at the kth moment.
进一步的，S4中，对于放电过程中，利用滤波算法结合最小二乘特征曲线间接得到放电段的直流内阻，对于放电过程中，由于电动汽车放电工况的复杂性，定义充电电流符号为负，放电电流符号为正，行驶过程中既存在电流大于0的正常放电过程，也存在电流小于0的制动能量回收，还存在其他工况引起的电流跳变，为保证放电内阻辨识的有效性，需要对电流跳变时刻进行滤波处理，经多次离线实验验证，选择0.1C(C指电池的(电流)倍率)的电流作为边界条件。将电流绝对值小于0.1C倍率的电流设置为0，则根据上式计算得到的内阻为无穷大，即不作参考。通过此方法滤波后，即可用充电阶段辨识直流内阻的方式辨识放电时刻的直流内阻。需注意的是，由于放电工况的变化复杂性，放电直流内阻仅作为安全预警的一个检测参考指标。Furthermore, in S4, during the discharge process, the DC internal resistance of the discharge section is indirectly obtained by using a filtering algorithm combined with the least square characteristic curve. During the discharge process, due to the complexity of the discharge conditions of electric vehicles, the charging current sign is defined as negative and the discharge current sign is defined as positive. During the driving process, there are both normal discharge processes with a current greater than 0 and braking energy recovery with a current less than 0. There are also current jumps caused by other conditions. In order to ensure the effectiveness of the identification of the discharge internal resistance, it is necessary to filter the current jump moment. After multiple offline experimental verifications, a current of 0.1C (C refers to the (current) rate of the battery) is selected as the boundary condition. If the current with an absolute value less than 0.1C rate is set to 0, the internal resistance calculated according to the above formula is infinite, that is, it is not used as a reference. After filtering by this method, the DC internal resistance at the discharge moment can be identified by identifying the DC internal resistance in the charging stage. It should be noted that due to the complexity of the changes in the discharge conditions, the discharge DC internal resistance is only used as a detection reference indicator for safety warning.
进一步的，S5中充电片段在时间维度上预警方法为：Furthermore, the warning method of the charging segment in S5 in the time dimension is:
首先，选取所有单体在70％-90％SOC区间的内阻：First, select the internal resistance of all monomers in the 70%-90% SOC range:
其中，j表示单体电池个数，i表示不同SOC的第i个采样点；Wherein, j represents the number of single cells, and i represents the i-th sampling point of different SOCs;
然后，将矩阵R的每一行进行标准化处理，其计算式为：Then, each row of the matrix R is standardized, and the calculation formula is:
式中ri表示R矩阵的行向量，Rin表示第i行第n列的矩阵元素，经过处理后可得到无量纲的R_normal矩阵：Whereri represents the row vector of the R matrix, andRin represents the matrix element in the i-th row and n-th column. After processing, the dimensionless R_normal matrix can be obtained:
计算R_normal行向量的均值和标准差Mi和Si：Compute the mean and standard deviationMi andSi of the row vector of R_normal:
式中n表示电池单体的个数；Where n represents the number of battery cells;
最后，求得不同采样点的变异系数Vi：Finally, the coefficient of variationVi of different sampling points is obtained:
对于变异系数，结果在0.15以内认为是各个电池单体一致性良好，当某一时刻的变异系数大于0.15则表明该时刻出现电池单体的不一致。For the coefficient of variation, if the result is within 0.15, it is considered that the consistency of each battery cell is good. When the coefficient of variation at a certain moment is greater than 0.15, it indicates that the battery cell is inconsistent at that moment.
进一步的，S5中充电片段在空间维度上预警方法为：Furthermore, the early warning method of the charging segment in S5 in the spatial dimension is:
首先，选取n个电池单体，m个采样时刻，建立矩阵Xij(i＝1,2,…,n；j＝1,2,…,m)，然后对矩阵列向量进行归一化，计算第j个采样点下第i个电池单体占该时刻的比重pij：First, select n battery cells and m sampling moments, establish a matrixXij (i = 1, 2, ..., n; j = 1, 2, ..., m), then normalize the matrix column vector, and calculatethe proportion pij of the i-th battery cell at the j-th sampling point at that moment:
然后，计算第j个时刻的熵值ej：Then, calculate the entropy value ej at the jth moment:
其中，根据熵值计算信息熵冗余度dj：in, Calculate the information entropy redundancy dj according to the entropy value:
dj＝1-ejdj =1-ej
对冗余度进行权值计算：Calculate the weight of redundancy:
利用每个单体计算所有j时刻下的总和可以得到各个电池单体的综合得分si：By calculating the sum of all j moments for each cell, we can get the comprehensive score si of each battery cell:
最后，将每个电池单体综合得分与所有单体得分的均值进行作差得到偏离度Δs，偏离度越大，反应该单体偏离大多数单体的程度越大，通过偏离的程度来判断该单体是否出现异常，经实验试错，偏离度的阈值定为4。Finally, the deviation Δs is obtained by subtracting the comprehensive score of each battery cell from the mean score of all cells. The larger the deviation, the greater the degree of deviation of the cell from the majority of cells. The degree of deviation is used to determine whether the cell is abnormal. After trial and error, the threshold of the deviation is set to 4.
进一步的，S3中提取30个充电片段起始时刻、运行过程中电流为零超过30min的时刻、充电结束并充分静置后的时刻的电压值。Furthermore, in S3, voltage values are extracted at the start time of 30 charging segments, at the time when the current is zero for more than 30 minutes during operation, and at the time when charging is completed and the battery is fully left to stand.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210913103.1ACN115469226B (en) | 2022-08-01 | 2022-08-01 | Real-time safety early warning method for power battery of electric automobile based on operation big data |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210913103.1ACN115469226B (en) | 2022-08-01 | 2022-08-01 | Real-time safety early warning method for power battery of electric automobile based on operation big data |
| Publication Number | Publication Date |
|---|---|
| CN115469226A CN115469226A (en) | 2022-12-13 |
| CN115469226Btrue CN115469226B (en) | 2024-08-20 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210913103.1AActiveCN115469226B (en) | 2022-08-01 | 2022-08-01 | Real-time safety early warning method for power battery of electric automobile based on operation big data |
| Country | Link |
|---|---|
| CN (1) | CN115469226B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116027212B (en)* | 2023-02-10 | 2025-09-02 | 上海电享信息科技有限公司 | Battery cell anomaly identification method, device and electronic equipment based on multi-method fusion |
| CN116142029A (en)* | 2023-04-07 | 2023-05-23 | 北京新能源汽车股份有限公司 | A power battery safety early warning method, device and vehicle |
| CN116466241B (en)* | 2023-05-06 | 2024-03-26 | 重庆标能瑞源储能技术研究院有限公司 | Thermal runaway positioning method for single battery |
| CN116520176A (en)* | 2023-05-06 | 2023-08-01 | 福建星云电子股份有限公司 | A battery internal resistance testing method, system, equipment and medium |
| CN120280521B (en)* | 2025-06-09 | 2025-09-09 | 杭州德海艾科能源科技有限公司 | Method, device, equipment and medium for analyzing abnormality of energy storage system of all-vanadium redox flow battery |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109669133A (en)* | 2019-01-18 | 2019-04-23 | 北京交通大学 | A kind of dynamic lithium battery lifetime data backstage mining analysis method |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105607009B (en)* | 2016-02-01 | 2018-05-01 | 深圳大学 | A kind of power battery SOC methods of estimation and system based on dynamic parameter model |
| CN106443474B (en)* | 2016-10-09 | 2019-03-26 | 北京理工大学 | A kind of electrokinetic cell system service life Decline traits quickly know method for distinguishing |
| CN107064815B (en)* | 2017-03-31 | 2019-09-20 | 惠州市蓝微新源技术有限公司 | A kind of internal resistance of cell calculation method |
| CN109061537B (en)* | 2018-08-23 | 2019-07-16 | 重庆大学 | Observer-based fault diagnosis method for electric vehicle lithium-ion battery sensor |
| CN111483469B (en)* | 2020-04-27 | 2021-08-03 | 湖南大学 | An Analysis and Testing Method for Fault Diagnosis of Electric Vehicle Vehicle Controller |
| CN111707951B (en)* | 2020-06-22 | 2021-04-06 | 北京理工大学 | Battery pack consistency evaluation method and system |
| CN112505550B (en)* | 2020-11-26 | 2022-06-07 | 重庆长安汽车股份有限公司 | A power battery monitoring and early warning method |
| CN112965001B (en)* | 2021-02-09 | 2024-06-21 | 重庆大学 | Power battery pack fault diagnosis method based on real vehicle data |
| CN113075554B (en)* | 2021-03-26 | 2022-07-05 | 国网浙江省电力有限公司电力科学研究院 | Lithium ion battery pack inconsistency identification method based on operation data |
| CN113253128B (en)* | 2021-05-12 | 2022-04-12 | 合肥国轩高科动力能源有限公司 | Battery system SOC consistency evaluation method and internal resistance consistency evaluation method |
| CN113300436A (en)* | 2021-06-11 | 2021-08-24 | 上海玫克生储能科技有限公司 | Dynamic management and control method for lithium battery energy storage system |
| CN113759265A (en)* | 2021-08-24 | 2021-12-07 | 珠海银隆电器有限公司 | Fault judgment method of power supply system and energy storage system |
| CN113777515A (en)* | 2021-09-13 | 2021-12-10 | 国网江西省电力有限公司供电服务管理中心 | A kind of electric vehicle charging safety warning method |
| CN114236393B (en)* | 2021-11-30 | 2023-07-07 | 上海瑞浦青创新能源有限公司 | Method and system for detecting battery abnormality on line based on big data |
| CN114325405A (en)* | 2021-12-31 | 2022-04-12 | 中国第一汽车股份有限公司 | Battery pack consistency analysis method, modeling method, device, equipment and medium |
| CN114464906A (en)* | 2022-02-10 | 2022-05-10 | 重庆金康动力新能源有限公司 | Power battery early warning method and device |
| CN114690040A (en)* | 2022-03-31 | 2022-07-01 | 日照职业技术学院 | Method for predicting optimal charging initial SOC of power battery of electric vehicle |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109669133A (en)* | 2019-01-18 | 2019-04-23 | 北京交通大学 | A kind of dynamic lithium battery lifetime data backstage mining analysis method |
| Title |
|---|
| 基于运行数据的异常电池诊断及实现;李欢;CNKI优秀硕士论文全文库;20181215;第7-48页* |
| 李欢.基于运行数据的异常电池诊断及实现.CNKI优秀硕士论文全文库.2018,第7-48页.* |
| Publication number | Publication date |
|---|---|
| CN115469226A (en) | 2022-12-13 |
| Publication | Publication Date | Title |
|---|---|---|
| CN115469226B (en) | Real-time safety early warning method for power battery of electric automobile based on operation big data | |
| CN113933732B (en) | New energy automobile power battery health state analysis method, system and storage medium | |
| CN114430080B (en) | A method for identifying abnormal self-discharge of power battery cells based on operating data | |
| CN110018425B (en) | Power battery fault diagnosis method and system | |
| CN111208439B (en) | Quantitative detection method for micro-short-circuit faults of series-connected lithium-ion battery packs | |
| CN107843853B (en) | Power battery pack series connection fault diagnosis method | |
| Wu et al. | A fault detection method of electric vehicle battery through Hausdorff distance and modified Z-score for real-world data | |
| CN111965559B (en) | On-line estimation method for SOH of lithium ion battery | |
| CN114559819B (en) | Electric automobile battery safety early warning method based on signal processing | |
| Shen et al. | Detection and quantitative diagnosis of micro-short-circuit faults in lithium-ion battery packs considering cell inconsistency | |
| CN115032542A (en) | A hybrid model-based prediction method for battery thermal runaway in energy storage systems | |
| CN113687251B (en) | Double-model-based lithium ion battery pack voltage abnormality fault diagnosis method | |
| CN115166555B (en) | Method and system for identifying abnormal internal resistance of battery module of battery big data platform | |
| CN117607704A (en) | Lithium ion battery pack micro-short circuit fault diagnosis method considering inconsistency | |
| CN112965001A (en) | Power battery pack fault diagnosis method based on real vehicle data | |
| CN117930027B (en) | A method, device and platform for detecting abnormal capacity of power battery | |
| CN116520194A (en) | A diagnostic method for internal short-circuit fault and capacity loss of lithium-ion batteries | |
| CN117783890A (en) | New energy automobile battery voltage fault diagnosis method based on operation data | |
| CN114646888A (en) | A method and system for evaluating the capacity decay of a power battery | |
| CN118425818A (en) | A power battery risk assessment method integrating health decline and consistency | |
| CN113533985A (en) | A method for identifying abnormal battery pack internal resistance module and its storage medium | |
| CN117805639A (en) | Method and system for judging micro-short circuit fault of lithium battery of electrochemical energy storage power station | |
| CN115792634A (en) | A battery cell voltage sampling fault identification method based on cloud online data | |
| CN105717458B (en) | A kind of on-line real-time measuremen method of the internal resistance of cell | |
| CN119064801A (en) | A battery status prediction method, system, device and medium |
| 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| OL01 | Intention to license declared | ||
| OL01 | Intention to license declared |