Battery internal resistance prediction method and device and electronic equipment

Abstract

The invention provides a battery internal resistance prediction method and device and electronic equipment, and relates to the technical field of battery detection. The method comprises the following steps: acquiring positive and negative electrode open-circuit potential curve data and production line data of a battery; performing feature extraction on the production line data to obtain initial feature data corresponding to each piece of production line data of each battery; according to the initial feature data and the positive and negative electrode open-circuit potential curve data, the influence degree of the battery internal resistance is screened to obtain a training set; taking the actually measured internal resistance of the battery after the capacity grading stage as a label value; and training a preset initial model by using the training set and the label value to obtain an internal resistance prediction model for predicting the internal resistance of the battery, and predicting the internal resistance of the battery to be detected by using the internal resistance prediction model. Therefore, the problems that the internal resistance of the battery in various charge states cannot be detected at the same time, the input data of a prediction model is complicated and the calculated amount is relatively large in the traditional detection or prediction of the internal resistance of the battery can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery detection, and in particular to a method, device and electronic equipment for predicting internal resistance of a battery. Background Art

[0002] With the increasing popularity of new energy vehicles, more and more users are buying new energy vehicles, and new energy vehicles are gradually becoming people's main means of transportation. As the main energy source for new energy vehicles, the performance of lithium-ion power batteries directly determines the user's driving experience. During the use of the battery, with continuous charging and discharging, the performance of the lithium-ion battery will decline, the internal resistance of the battery will become larger and larger, and the power will become worse and worse. Moreover, with different degrees of use and different driving habits, the internal resistance of the battery of the same vehicle will also change differently. Therefore, in order to understand the health of the vehicle battery's internal resistance when it leaves the factory, so as to facilitate real-time monitoring of the battery's internal resistance in subsequent use, the battery can also be preliminarily screened at the production line end.

[0003] When the existing battery production line performs internal resistance testing on the production line battery, it usually only tests the internal resistance of the battery at a certain SOC, but the battery management system (BMS) needs to know the internal resistance at multiple SOCs during actual battery use. In addition, since there are many factors that affect the internal resistance of the battery, and the degree of influence of each factor on the internal resistance is different, the existing method of predicting the internal resistance through a regression model, which inputs all the influencing factors into the model for internal resistance prediction, has the problem of complex data and excessive calculation. Summary of the invention

[0004] In view of this, the purpose of the embodiments of the present application is to provide a battery internal resistance prediction method, device and electronic device, which can improve the problems of traditional battery internal resistance detection or prediction, such as the inability to simultaneously detect the battery internal resistance under multiple charge states, the complex input data of the prediction model and the large amount of calculation.

[0005] In order to achieve the above technical objectives, the technical solutions adopted in this application are as follows:

[0006] In a first aspect, an embodiment of the present application provides a method for predicting internal resistance of a battery, the method comprising:

[0007] Obtaining positive and negative electrode open circuit potential curve data and production line data of the initial sample battery, wherein the production line data characterizes the parameters that affect the internal resistance of the battery during the production and internal resistance detection stages of the initial sample battery;

[0008] Extracting features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries;

[0009] Screening is performed according to the influence of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set;

[0010] The measured internal resistance of each of the initial sample batteries after the capacity division stage is used as a label value;

[0011] The preset initial model is trained using the training set and the corresponding label value to obtain an internal resistance prediction model for battery internal resistance prediction, and the internal resistance of the battery to be tested is predicted using the internal resistance prediction model.

[0012] In combination with the first aspect, in some optional implementations, the method further includes:

[0013] The initial characteristic data and the positive and negative electrode open circuit potential data are used as important characteristic data, and preset labels are assigned to the important characteristic data, wherein the preset labels represent the influence of each of the important characteristic data on the internal resistance of the battery;

[0014] The initial characteristic data and the positive and negative electrode open circuit potential curve data are screened according to their influence on the internal resistance of the battery to obtain a training set, including:

[0015] Sorting each of the important feature data items according to the preset label, and randomly arranging each of the sorted important feature data items to obtain a reference feature set corresponding to each of the important feature data items;

[0016] Inputting the important feature data and each group of the reference feature sets into the preset initial model to perform internal resistance prediction, and obtaining a first prediction result corresponding to the important feature data and a second prediction result corresponding to each group of the reference feature sets;

[0017] Acquire first measured internal resistance data corresponding to the important characteristic data, and second measured internal resistance data corresponding to each group of the reference characteristic sets;

[0018] The important feature data are screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data, to obtain the training set.

[0019] In combination with the first aspect, in some optional implementations, the important feature data is screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data to obtain the training set, including:

[0020] According to the first error and each of the second errors, an importance parameter of each of the important feature data is determined, wherein the importance parameter represents the strength of the correlation between each of the important feature data and the preset initial model:

[0021]

[0022] In the formula, PI i represents the importance parameter corresponding to the i-th reference feature, E i represents the second error corresponding to the i-th reference feature, and E represents the first error;

[0023] When the importance parameter is greater than a preset threshold, the important feature data corresponding to the current importance parameter is deleted from all the important feature data, and the set of the remaining important feature data in the important feature data is used as the training set.

[0024] In combination with the first aspect, in some optional implementations, the method for obtaining the measured internal resistance includes:

[0025] Obtaining a total voltage difference and a total current difference of each of the initial sample batteries at two adjacent moments at a preset SOC stage after the capacity division stage;

[0026] According to the total voltage difference and the total current difference, the measured internal resistance corresponding to each of the initial sample batteries is calculated by the following formula:

[0027]

[0028] Where R represents the measured internal resistance, U K1 Represents the total voltage difference at the previous moment, U K2 Represents the total voltage difference at the next moment, I K1 Represents the total current difference at the previous moment, I K2 Indicates the total current difference at the next moment.

[0029] In combination with the first aspect, in some optional implementations, the method further includes:

[0030] In the process of predicting the internal resistance of the battery to be tested by using the internal resistance prediction model, some of the batteries to be tested are extracted as calibration sample batteries, and the internal resistance of the calibration sample batteries is measured to obtain the calibrated measured internal resistance. When the error between the calibrated measured internal resistance and the internal resistance of the battery to be tested predicted by the internal resistance prediction model is greater than a preset error, a training set obtained based on the positive and negative open-circuit potential curve data of the calibration sample resistor and the production line data is used as a new training set, and the calibrated measured internal resistance is used as a new label value. The internal resistance prediction model is trained using the new training set and the new label value to update the internal resistance prediction model.

[0031] In combination with the first aspect, in some optional implementations, before screening the influence of each item of the initial feature data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set, the method further includes:

[0032] Preprocessing each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data to obtain the corresponding preprocessed initial characteristic data and the positive and negative electrode open circuit potential curve data;

[0033] Then, the influence of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance is screened to obtain a training set including:

[0034] According to the influence of the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data are screened to obtain the training set.

[0035] In conjunction with the first aspect, in some optional implementations, the initial feature data includes at least one of the following parameters:

[0036] The preset process parameters in the battery production process, the insulation withstand voltage test curve parameters of the uninjected battery, the dQdv curve parameters calculated according to the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process, the segmentation mechanism parameters after segmenting the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process based on the dQdv curve, and the dQdt curve parameters calculated based on the battery temperature curve of the formation process and the battery temperature curve of the charging section of the capacity division process.

[0037] In combination with the first aspect, in some optional embodiments, the preset process parameters include at least one of the slurry viscosity, coating and drying temperature, rolling thickness, pole piece turnover time and injection volume in the battery production process.

[0038] In combination with the first aspect, in some optional implementations, the preset error is 5%.

[0039] In a second aspect, an embodiment of the present application further provides a battery internal resistance prediction device, the device comprising:

[0040] A first acquisition unit is used to acquire the positive and negative open circuit potential curve data of the initial sample battery and the production line data of multiple batteries, wherein the production line data represents the parameters that affect the battery internal resistance during the production and internal resistance detection stages of the initial sample battery;

[0041] A feature extraction unit, configured to extract features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries;

[0042] A screening unit, used to screen the influence of each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set;

[0043] A label value extraction unit, which uses the measured internal resistance of each of the initial sample batteries after the capacity division stage as a label value;

[0044] A training unit trains the preset initial model using the training set and the corresponding flag value to obtain an internal resistance prediction model for battery internal resistance prediction, so as to use the internal resistance prediction model to predict the internal resistance of the battery to be tested.

[0045] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a processor and a memory coupled to each other, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the electronic device executes the above method.

[0046] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed on a computer, the computer executes the above method.

[0047] The invention adopting the above technical solution has the following advantages:

[0048] In the technical solution provided by the present application, firstly, the positive and negative electrode open circuit potential curve data and production line data of the initial sample battery are obtained; then, the production line data is feature extracted to obtain the initial feature data corresponding to each item of the production line data of each initial sample battery; then, according to the degree of influence of each item of the initial feature data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, a screening is performed to obtain a training set; the measured internal resistance of each of the initial sample batteries after the capacity division stage is used as a label value; then, the preset initial model is trained using the training set and the corresponding label value to obtain an internal resistance prediction model for battery internal resistance prediction; after obtaining the internal resistance prediction model, the internal resistance of the battery to be tested can be predicted by the internal resistance prediction model. In this way, the problems of traditional battery internal resistance detection or prediction, the inability to simultaneously detect the internal resistance of batteries under multiple charge states, and the complicated input data of the prediction model and the large amount of calculation can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The present application may be further described by the non-limiting embodiments given in the accompanying drawings. It should be understood that the following drawings only illustrate certain embodiments of the present application and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings may be obtained based on these drawings without creative effort.

[0050] Figure 1 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0051] Figure 2 A schematic diagram of the flow chart of the battery internal resistance prediction method provided in an embodiment of the present application.

[0052] Figure 3 A schematic diagram of the important feature data screening process provided in an embodiment of the present application.

[0053] Figure 4 A schematic diagram of the internal resistance prediction model update process provided in an embodiment of the present application.

[0054] Figure 5 This is a structural block diagram of a battery internal resistance prediction device provided in an embodiment of the present application.

[0055] Icon: 100 - electronic device; 101 - processor; 102 - memory; 200 - battery internal resistance prediction device; 210 - first acquisition unit; 220 - feature extraction unit; 230 - screening unit; 240 - label value extraction unit; 250 - training unit. DETAILED DESCRIPTION

[0056] The present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that in the drawings or descriptions, similar or identical parts use the same figure numbers, and the implementation methods not shown or described in the drawings are forms known to ordinary technicians in the relevant technical field. In the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0057] Please refer to Figure 1 , an embodiment of the present application provides an electronic device 100 that may include a processor 101 and a memory 102. The memory 102 stores a computer program, and when the computer program is executed by the processor 101, the electronic device 100 can perform the corresponding steps in the following battery internal resistance prediction method.

[0058] In this embodiment, the electronic device 100 can be a personal computer, a laptop computer, a cloud server, etc., which is used to obtain the positive and negative electrode open circuit potential curve data and production line data of the initial sample battery, wherein the production line data characterizes the parameters that affect the internal resistance of the battery during the production and internal resistance detection stages of the initial sample battery; and the production line data is feature extracted to obtain the initial feature data corresponding to each item of the production line data of each initial sample battery. Then, according to the different degrees of influence of each item of initial feature data and positive and negative electrode open circuit potential curve data on the internal resistance of the battery, screening is performed to obtain a training set. The measured internal resistance of each of the initial sample batteries after the capacity division stage is used as a label value; the preset initial model is trained using the training set and the corresponding label value to obtain a battery internal resistance prediction model. Finally, the battery internal resistance is predicted in real time through this model.

[0059] Please refer to Figure 2 The present application also provides a battery internal resistance prediction method, which can be applied to the above electronic device 100, and each step in the method is executed or implemented by the electronic device 100. The battery internal resistance prediction method may include the following steps:

[0060] Step 110, obtaining the positive and negative open circuit potential curve data of the initial sample battery and the production line data of multiple batteries, wherein the production line data represents the parameters that affect the internal resistance of the initial sample battery during the production and internal resistance detection stages;

[0061] Step 120, extracting features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries;

[0062] Step 130, screening the influence of each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set;

[0063] Step 140, taking the measured internal resistance of each of the initial sample batteries after the capacity division stage as a label value;

[0064] Step 150, using the training set and the corresponding label value to train the preset initial model to obtain an internal resistance prediction model for battery internal resistance prediction, and using the internal resistance prediction model to predict the internal resistance of the battery to be tested.

[0065] In the above-mentioned implementation, firstly, the positive and negative electrode open circuit potential curve data and production line data of the initial sample battery are obtained; then, the production line data is feature extracted to obtain the initial feature data corresponding to each item of the production line data of each initial sample battery; then, according to the influence of each item of the initial feature data and the positive and negative electrode open circuit potential curve data on the battery internal resistance, screening is performed to obtain a training set; then, the measured internal resistance of each initial sample battery after the capacity division stage is used as a label value; the preset initial model is trained using the training set and the corresponding label value to obtain an internal resistance prediction model for battery internal resistance prediction; after obtaining the internal resistance prediction model, the internal resistance of the battery to be tested can be predicted by the internal resistance prediction model. In this way, the problems of traditional battery internal resistance detection or prediction, the inability to simultaneously detect the battery internal resistance under multiple charge states, and the complicated input data and large amount of calculation of the prediction model can be improved.

[0066] The following will describe in detail the steps of the battery internal resistance prediction method, as follows:

[0067] In step 110, the positive and negative open circuit potential curve data and production line data of the initial sample battery are obtained, wherein the production line data characterizes the parameters that affect the internal resistance of the battery during the production and internal resistance detection stages of the initial sample battery. In this embodiment, the positive and negative open circuit potential curves of the initial sample battery can be obtained by taking a piece of negative electrode from the battery production line, cutting it into a round piece, and assembling it with a diaphragm, a lithium sheet, an electrolyte, and a shell into a button battery. Then, a small current full charge test is performed on the button battery to obtain the negative electrode open circuit potential curve. In the same way, take a positive electrode from the battery production line and test it to get the positive open circuit potential curve. Wherein, x and y are the relative lithium ion contents of the negative electrode and the positive electrode, respectively, ranging from 0 to 1, and the unit is 1.

[0068] In step 120, feature extraction is performed on the production line data to obtain initial feature data corresponding to each item of the production line data of each initial sample battery, the initial feature data including preset process parameters in the battery production process, insulation withstand voltage detection curve parameters of the uninjected battery, dQdv curve parameters calculated based on the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process, segmentation mechanism parameters after segmenting the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process based on the dQdv curve, and dQdt curve parameters calculated based on the battery temperature curve of the formation process and the battery temperature curve of the charging section of the capacity division process.

[0069] The preset process parameters in the battery production process can obtain the battery production data through the commonly used execution system or production equipment in the battery production process, which may include the battery slurry viscosity, coating and drying temperature, rolling thickness, pole piece turnover time, injection volume, etc. In practical applications, the aforementioned process parameters are usually single values ​​and can be directly used as inputs for subsequent models.

[0070] The insulation withstand voltage test curve parameters of the unfilled battery can be obtained by extracting the characteristics of the voltage curve during the insulation withstand voltage test of the unfilled battery, and then taking the average voltage in the current excitation stage as the insulation withstand voltage test curve parameters of the unfilled battery.

[0071] The dQdv curve parameters are obtained as follows:

[0072] First, the characteristic parameters of the dQdV curve (differential capacity curve) are calculated through the voltage curve of the formation process and the voltage curve of the charging stage of the capacity division process. The calculation of the dQdV curve adopts the point method, and the formula is as follows:

[0073]

[0074] In the formula, I represents the battery current, Δt represents the time step of the data curve, and ΔV represents the selected voltage window width. Indicates that the voltage is The number of data points in the interval. i It is usually an integer multiple of ΔV, and the upper and lower limits are selected according to the maximum and minimum values ​​of the actual voltage curve. It can be understood that the dQdV curve usually contains multiple peaks, and the peaks of the dQdV curve usually correspond to the platform area on the voltage curve. Through the peak extraction algorithm, multiple peaks on the dQdV curve can be extracted, and the voltage corresponding to each peak can be used as the dQdV parameter of the battery, that is, [(dQdV p1 ,V p1 ),(dQdV p2 ,V p2 ),(dQdV p3 ,V p3 )……]. The specific number of peaks is set according to the geometric characteristics actually shown by the dQdV curve. In practical applications, small peaks can be filtered out and multiple main peaks can be left by limiting the minimum relative height, minimum absolute height, minimum width and other indicators of the peaks.

[0075] In this embodiment, the battery production process (i.e., production line) mainly includes stages / processes / steps such as homogenization, coating, slitting, formation, and capacity separation. The battery is activated in the formation stage, and then the battery capacity is tested in the capacity separation stage to screen out abnormal batteries, thereby ensuring the quality of the battery.

[0076] The acquisition of segmented mechanism parameters is as follows: first, based on the dQdV curve, the voltage curve is segmented. The segmented mechanism parameters are the characteristic parameters of each segment of the voltage curve after segmentation. It can be understood that the voltage curve usually contains a large number of platforms and inflection points, and the voltage curve is divided into multiple segments based on this, each segment contains different mechanism information. Then, according to the peak position (V p1 ,V p2 ,V p3 ……), the voltage curve is divided into multiple segments. Among them, the voltage value is between [0,V p1 ) is a data point, and the voltage value is [V p1 ,V p2 ) is a segment, and so on. For each segment on each voltage curve, a set of characteristic parameters corresponding to the segment of the voltage curve (i.e., segmented mechanism parameters) are calculated, including voltage mean, voltage standard deviation, voltage skewness, etc.

[0077] In this embodiment, the dQdt curve parameters are obtained as follows: the battery temperature curve of the formation process and the battery temperature curve of the capacity separation process can be calculated to obtain the characteristic parameters of the dQdT curve as the dQdT parameters. Specifically, the calculation of the dQdT curve adopts point method, and the formula is as follows:

[0078]

[0079] In the formula, I represents the battery current, Δt represents the time step of the data curve, and ΔT represents the selected temperature window width. Indicates that the temperature is The number of data points in the interval. i It is usually an integer multiple of ΔT, and the upper and lower limits are selected according to the maximum and minimum values ​​of the actual temperature curve. Then, the characteristic parameters of the temperature curve, namely the dQdT parameters, are obtained by calculating the mean, change rate, peak value, etc. of the dQdT curve.

[0080] After obtaining the above initial feature data, each initial feature data is respectively converted into K1, K2, K3...K n express.

[0081] Between step 120 and step 130, the method may further include:

[0082] Preprocessing each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data to obtain the corresponding preprocessed initial characteristic data and the positive and negative electrode open circuit potential curve data;

[0083] Then, the influence of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance is screened to obtain a training set including:

[0084] According to the influence of the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data are screened to obtain the training set.

[0085] In this embodiment, the initial characteristic data and the positive and negative electrode open circuit potential data are used as important characteristic data, and the preprocessing of the important characteristic data can be performed by performing Kalman filtering on the important characteristic data.

[0086] In this way, data with excessive noise and deviation is filtered out, and important feature data is preliminarily screened, making the data more accurate, improving the accuracy of subsequent model construction, and making the final prediction results more reliable.

[0087] In step 130, a preset label is assigned to the important characteristic data, wherein the preset label represents the influence degree of each important characteristic data on the internal resistance of the battery;

[0088] The initial characteristic data and the positive and negative electrode open circuit potential curve data are screened according to their influence on the internal resistance of the battery to obtain a training set, including:

[0089] Sorting each of the important feature data items according to the preset label, and randomly arranging each of the sorted important feature data items to obtain a reference feature set corresponding to each of the important feature data items;

[0090] Inputting the important feature data and each group of the reference feature sets into the preset initial model to perform internal resistance prediction, and obtaining a first prediction result corresponding to the important feature data and a second prediction result corresponding to each group of the reference feature sets;

[0091] Acquire first measured internal resistance data corresponding to the important characteristic data, and second measured internal resistance data corresponding to each group of the reference characteristic sets;

[0092] The important feature data are screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data, to obtain the training set.

[0093] In this embodiment, the preset label can be obtained by performing importance analysis on various important characteristic data of the battery according to Pearson correlation analysis to obtain important characteristic sequences of various important characteristic data of the battery as the preset label.

[0094] In this embodiment, the important feature data is screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data to obtain the training set, which may include:

[0095] According to the first error and each of the second errors, an importance parameter of each of the important feature data is determined, wherein the importance parameter represents the strength of the correlation between each of the feature data and the preset initial model:

[0096]

[0097] In the formula, PI i represents the importance parameter corresponding to the i-th reference feature, E i represents the second error corresponding to the i-th reference feature, and E represents the first error;

[0098] When the importance parameter is greater than a preset threshold, the important feature data corresponding to the current importance parameter is deleted from all the important feature data, and the set of the remaining important feature data in the important feature data is used as the training set.

[0099] In this embodiment, the battery data characteristics (i.e., important characteristic data) K1, K2, K3, ..., K n , sort these feature data according to preset labels, and then refer to Figure 3 , select one of the important feature data (K1, K2, K3...K n Any one of the items is K i (1≤i≤n)), shuffle the order of the data therein, and then input the shuffled important feature data matrix (i.e., reference feature set) and the important feature data matrix before shuffling into the preset initial model respectively, to obtain the first prediction result corresponding to the important feature data and the second prediction result corresponding to the reference feature set, and then obtain the measured internal resistance data calculated during the acquisition of each important feature data, i.e., the first measured internal resistance data and the second measured internal resistance data. Then, respectively calculate the first error E between the first prediction result and the first measured data, and the second error E between the second prediction result and the second measured internal resistance data. i Then the importance parameter representing the correlation between each important feature data and the preset initial model is calculated by the following formula:

[0100]

[0101] In the formula, PI i represents the importance parameter corresponding to the i-th reference feature, E irepresents the second error corresponding to the i-th reference feature, and E represents the first error.

[0102] Among them, PI i The smaller the value, the higher the correlation between the feature and the model. On the contrary, the smaller the correlation between the feature and the model. i When it is greater than the preset threshold (which can be flexibly set according to user needs), it indicates that the important feature data is irrelevant to the construction of the model and will not affect the prediction results of the model. The important feature data can be deleted to reduce the calculation amount and workload of the model. Then select the important feature data K that has not been disrupted j (1≤j≤n, j≠i) and repeat the above steps to get a new PI i Numerical values, filter important characteristic data until K1, K2, K3...K n Each important feature data in the dataset was individually shuffled and filtered.

[0103] In this way, by screening various important feature data, a lot of unnecessary important feature data is removed, the features that significantly affect the model prediction results are retained, and these retained features can be sorted by importance. This not only improves the accuracy of internal resistance prediction, but also reduces a considerable amount of calculation for subsequent model building and saves calculation time.

[0104] In step 130, the preset initial model may be a feature regression model based on LSTM, and the preset initial model may be expressed as Among them, j represents the number of hidden layers, f represents the activation function, i represents the number of features, W represents the hidden layer weight, K represents the feature data, H represents the hidden state, and b h represents the hidden layer bias, b0 represents the output layer bias, and R represents the measured internal resistance corresponding to the training set.

[0105] In step 140, the measured internal resistance of each of the initial sample batteries after the capacity division stage is used as a label value. In practical applications, the training of the model requires the training set and the measured internal resistance (i.e., label value) corresponding to the training set as data for model training. In this embodiment, the total voltage difference and total current difference of each of the initial sample batteries at two adjacent moments at the preset SOC stage after the capacity division stage are obtained, specifically: the voltage difference and current difference at the constant current section are switched at 20%, 50%, and 80% SOC after the capacity division stage of the automobile battery, and the total voltage difference and total current difference at the moment before the switching are obtained, as well as the total voltage difference and total current difference at the moment after the switching. Then the measured internal resistance is calculated by the following formula:

[0106]

[0107] In the formula, R represents internal resistance, U K1Represents the total voltage difference at the previous moment, U K2 Represents the total voltage difference at the next moment, I K1 Represents the total current difference at the previous moment, I K2 Indicates the total current difference at the next moment.

[0108] In this embodiment, the polarization internal resistance R1R2 of the battery can also be used as one of the output data sets of the model. The polarization internal resistance can be obtained at the production line end, by performing HPPC tests on the batteries after capacity division and collecting test data, and calculating the polarization internal resistance of the battery using a second-order RC equivalent circuit model, and using the polarization internal resistance as the output data set of the model to correct the model training.

[0109] In this way, after the model training is corrected by polarization internal resistance, the prediction accuracy of polarization internal resistance will also be improved. As one of the representations of the internal characteristics of the battery, predicting the polarization internal resistance of a batch of batteries on the production line can also help battery manufacturers to conduct preliminary screening of unhealthy batteries at the production line end.

[0110] In the model training process, the Bayesian-LSTM optimization method is used to use the probability proxy model to proxy the probability model of the unknown target function. By iteratively increasing the amount of information, the prior probability is continuously corrected, so that the proxy model is more accurate. In the model training, the feature neural network model is corrected by continuously inputting the filtered feature data and the calculated internal resistance data, so that the neural network model is more accurate. According to the posterior distribution, the acquisition function selects the most "potential" sample point from the candidate set by sampling in the area where the global optimal solution is most likely to appear and the area that has not been sampled, so that the loss function F is minimized, that is, the predicted internal resistance error is minimized. The definition of the loss function F is as follows:

[0111]

[0112] Where Y pre represents the predicted value of the model, Y real represents the calculated measured internal resistance, and n represents the number of samples of the input characteristic data.

[0113] Finally, the internal resistance prediction model is obtained according to the maximum number of iterations of the user-defined model training or the convergence of the loss function.

[0114] In practical applications, the accuracy index defined by the user can also be used to determine whether the results of model training meet the user's expectations, so that the user can determine whether repeated training is needed. The accuracy index can use two error indicators, relative error (RelativeError) and absolute error (Absolute Error), to evaluate the model. The calculation formulas for the two are as follows:

[0115]

[0116] AE=|y pre -y real |

[0117] In this way, the predicted DCR (discharge resistance) of the resistor to be tested, that is, the aforementioned internal resistance, is used as one of the basis for judging the health status of the vehicle battery. Batteries with excessively large DCR are removed or returned to the factory for processing, thereby achieving preliminary screening of the battery production line.

[0118] As an optional implementation, the method may further include:

[0119] Obtaining the total voltage difference and total current difference of each of the initial sample batteries at two adjacent moments at a preset SOC stage (i.e., at the constant current section switching at 20%, 50%, and 80% SOC after the car battery capacity division stage);

[0120] According to the total voltage difference and the total current difference, the measured internal resistance of each corresponding initial sample battery is calculated by the following formula:

[0121]

[0122] Where R represents the measured internal resistance, U K1 Represents the total voltage difference at the previous moment, U K2 Represents the total voltage difference at the next moment, I K1 Represents the total current difference at the previous moment, I K2 Indicates the total current difference at the next moment;

[0123] In the process of predicting the internal resistance of the battery to be tested by using the internal resistance prediction model, some of the batteries to be tested are extracted as calibration sample batteries, and the internal resistance of the calibration sample batteries is measured to obtain the calibrated measured internal resistance;

[0124] It should be understood that a small number of batteries can be selected as calibration sample batteries every certain number of batteries to be tested. The number of batteries to be tested is generally much larger than the calibration sample batteries. The method of measuring the calibration measured internal resistance is consistent with the above-mentioned method of calculating the measured internal resistance corresponding to each initial sample battery.

[0125] When the error between the calibrated measured internal resistance and the internal resistance of the battery to be tested predicted by the internal resistance prediction model is greater than a preset error, the training set obtained based on the open-circuit potential curve data of the positive and negative electrodes of the calibration sample battery and the production line data is used as a new training set, and the calibrated measured internal resistance is used as a new label value. The internal resistance prediction model is trained using the new training set and the new label value to update the internal resistance prediction model.

[0126] In this embodiment, refer to Figure 4 , the time difference between the previous moment K1 and the next moment K2 can be 10 seconds. Within these 10 seconds, the total resistance (i.e., internal resistance R) represents the sum of the ohmic internal resistance and polarization internal resistance of the battery, which can fully reflect the resistance characteristics of the battery. The prediction results obtained by bringing the training set obtained from the real-time collected positive and negative electrode open circuit potential curve data and production line data into the model training are compared with the measured internal resistance calculated by the above formula after capacity division, and the data with an error greater than 5% is brought into the aforementioned internal resistance prediction model for iterative training, so that the model can continuously learn data that has not been learned and adapt to more battery conditions.

[0127] Please refer to Figure 5 The present application also provides a battery internal resistance prediction device 200, which includes at least one software function module that can be stored in the memory 102 in the form of software or firmware or fixed in the operating system (OS) of the electronic device 100. The processor 101 is used to execute the executable modules stored in the memory 102, such as the software function modules and computer programs included in the battery internal resistance prediction device 200.

[0128] The battery internal resistance prediction device 200 includes a first acquisition unit 210, a feature extraction unit 220, a screening unit 230, a label value extraction unit 240 and a training unit 250. The functions of each unit may be as follows:

[0129] A first acquisition unit 210 is used to acquire the positive and negative open circuit potential curve data of the initial sample battery and the production line data of multiple batteries, wherein the production line data represents the parameters of the initial sample that affect the battery internal resistance during the production and internal resistance detection stages;

[0130] A feature extraction unit 220 is used to extract features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries;

[0131] A screening unit 230 is used to screen the influence of each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set;

[0132] The label value extraction unit 240 uses the measured internal resistance of each of the initial sample batteries after the capacity division stage as a label value;

[0133] The training unit 250 is used to train the preset initial model using the training set and the corresponding flag value to obtain an internal resistance prediction model for battery internal resistance prediction, so as to use the internal resistance prediction model to predict the internal resistance of the battery to be tested.

[0134] Optionally, the battery internal resistance prediction device 200 further includes:

[0135] a label unit, taking the initial characteristic data and the positive and negative electrode open circuit potential data as important characteristic data, and assigning a preset label to the important characteristic data, wherein the preset label represents the influence degree of each important characteristic data on the internal resistance of the battery;

[0136] The screening unit 230 is further used for:

[0137] Sorting each of the important feature data items according to the preset label, and randomly arranging each of the sorted important feature data items to obtain a reference feature set corresponding to each of the important feature data items;

[0138] Inputting the important feature data and each group of the reference feature sets into the preset initial model to perform internal resistance prediction, and obtaining a first prediction result corresponding to the important feature data and a second prediction result corresponding to each group of the reference feature sets;

[0139] Acquire first measured internal resistance data corresponding to the important characteristic data, and second measured internal resistance data corresponding to each group of the reference characteristic sets;

[0140] The important feature data are screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data, to obtain the training set.

[0141] Optionally, the screening unit 230 is further used for:

[0142] According to the first error and each of the second errors, an importance parameter of each of the important feature data is determined, wherein the importance parameter represents the strength of the correlation between each of the important feature data and the preset initial model:

[0143]

[0144] In the formula, PI i represents the importance parameter corresponding to the i-th reference feature, E i represents the second error corresponding to the i-th reference feature, and E represents the first error;

[0145] When the importance parameter is greater than a preset threshold, the important feature data corresponding to the current importance parameter is deleted from all the important feature data, and the set of the remaining important feature data in the important feature data is used as the training set.

[0146] Optionally, the battery internal resistance prediction device 200 further includes:

[0147] A third acquisition unit, for acquiring a total voltage difference and a total current difference of each of the initial sample batteries at two adjacent moments at a preset SOC stage after the capacity division stage;

[0148] The internal resistance calculation unit calculates the measured internal resistance corresponding to each of the initial sample batteries according to the total voltage difference and the total current difference by using the following formula.

[0149]

[0150] Where R represents the measured internal resistance, U K1 Represents the total voltage difference at the previous moment, U K2 Represents the total voltage difference at the next moment, I K1 Represents the total current difference at the previous moment, I K2 Indicates the total current difference at the next moment.

[0151] The updating unit uses the internal resistance prediction model to predict the internal resistance of the battery to be tested, extracts some of the batteries to be tested as calibration sample batteries, measures the internal resistance of the calibration sample batteries, and obtains the calibration measured internal resistance. When the error between the calibration measured internal resistance and the internal resistance of the battery to be tested predicted by the internal resistance prediction model is greater than a preset error, a training set obtained based on the open-circuit potential curve data of the positive and negative electrodes of the calibration sample battery and the production line data is used as a new training set, and the calibration measured internal resistance is used as the corresponding new label value. The internal resistance prediction model is trained by the new training set and the new label value to update the internal resistance prediction model.

[0152] Optionally, the battery internal resistance prediction device 200 further includes:

[0153] A preprocessing unit preprocesses each of the initial characteristic data and the positive and negative electrode open circuit potential curve data to obtain the corresponding preprocessed initial characteristic data and the positive and negative electrode open circuit potential curve data;

[0154] The screening unit 230 is further used for:

[0155] According to the influence of the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data are screened to obtain the training set.

[0156] Optionally, the initial feature data includes at least one of the following parameters:

[0157] The preset process parameters in the battery production process, the insulation withstand voltage test curve parameters of the uninjected battery, the dQdv curve parameters calculated according to the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process, the segmentation mechanism parameters after segmenting the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process based on the dQdv curve, and the dQdt curve parameters calculated based on the battery temperature curve of the formation process and the battery temperature curve of the charging section of the capacity division process.

[0158] Optionally, the preset process parameters include at least one of the slurry viscosity, coating and drying temperature, rolling thickness, pole piece turnover time and injection volume in the battery production process.

[0159] Optionally, the preset error is 5%.

[0160] In this embodiment, the processor 101 may be an integrated circuit chip having signal processing capabilities. The processor 101 may be a general-purpose processor. For example, the processor 101 may be a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and may implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application.

[0161] The memory 102 may be, but is not limited to, a random access memory, a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, etc. In this embodiment, the memory 102 may be used to store production line data, initial feature data, important feature data, training sets, preset initial models, internal resistance prediction models, real-time charging parameters, internal resistance of the battery to be tested, preset labels, preset thresholds, preset errors, etc. Of course, the memory 102 may also be used to store programs, and the processor 101 executes the program after receiving the execution instruction.

[0162] Understandably, Figure 1 The structure of the electronic device 100 shown in FIG. 1 is only a schematic diagram of a structure. The electronic device 100 may also include Figure 1 More components are shown. Figure 1 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0163] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the electronic device 100 described above can refer to the corresponding process of each step in the aforementioned method, and will not be elaborated herein.

[0164] The embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed on a computer, the computer executes the battery internal resistance prediction method as described in the above embodiment.

[0165] Through the description of the above implementation methods, technical personnel in this field can clearly understand that the present application can be implemented by hardware, and can also be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0166] In summary, the embodiments of the present application provide a method, device, electronic device and storage medium for predicting the internal resistance of a battery. In the present technical solution, the positive and negative electrode open circuit potential curve data and the production line data of the initial sample battery are obtained; then, the production line data is feature extracted to obtain the initial feature data corresponding to each item of the production line data of each initial sample battery; then, according to the degree of influence of each item of the initial feature data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, screening is performed to obtain a training set; the measured internal resistance of each initial sample battery after the capacity division stage is used as a label value; the preset initial model is trained using the training set and the corresponding label value to obtain an internal resistance prediction model for battery internal resistance prediction; after the internal resistance prediction model is obtained, the internal resistance of the battery to be tested can be predicted by the internal resistance prediction model. In this way, the problems of traditional battery internal resistance detection or prediction, the inability to simultaneously detect the internal resistance of the battery under multiple charge states, and the complicated input data of the prediction model and the large amount of calculation can be improved.

[0167] In the embodiments provided in the present application, it should be understood that the disclosed device, system and method can also be implemented in other ways. The above-described device, system and method embodiments are merely schematic, for example, the flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a part of a module, a program segment or a code, and a part of the module, program segment or code contains one or more executable instructions for implementing the specified logical function. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. In addition, each functional module in each embodiment of the present application can be integrated together to form an independent part, or each module can exist separately, or two or more modules can be integrated to form an independent part.

[0168] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting battery internal resistance, characterized in that: The method comprises: Obtaining positive and negative electrode open circuit potential curve data and production line data of the initial sample battery, wherein the production line data characterizes the parameters that affect the internal resistance of the battery during the production and internal resistance detection stages of the initial sample battery; Extracting features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries; Screening is performed according to the influence of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set; The measured internal resistance of each of the initial sample batteries after the capacity division stage is used as a label value; The preset initial model is trained using the training set and the corresponding label value to obtain an internal resistance prediction model for battery internal resistance prediction, and the internal resistance of the battery to be tested is predicted using the internal resistance prediction model.

2. The method according to claim 1, characterized in that The method further comprises: The initial characteristic data and the positive and negative electrode open circuit potential data are used as important characteristic data, and preset labels are assigned to the important characteristic data, wherein the preset labels represent the influence of each of the important characteristic data on the internal resistance of the battery; The initial characteristic data and the positive and negative electrode open circuit potential curve data are screened according to their influence on the internal resistance of the battery to obtain a training set, including: Sorting each of the important feature data items according to the preset label, and randomly arranging each of the sorted important feature data items to obtain a reference feature set corresponding to each of the important feature data items; Inputting the important feature data and each group of the reference feature sets into the preset initial model to perform internal resistance prediction, and obtaining a first prediction result corresponding to the important feature data and a second prediction result corresponding to each group of the reference feature sets; Acquire first measured internal resistance data corresponding to the important characteristic data, and second measured internal resistance data corresponding to each group of the reference characteristic sets; The important feature data are screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data, to obtain the training set.

3. The method according to claim 2, characterized in that The important feature data are screened according to a first error between the first prediction result and the first measured internal resistance data, and a second error between each second prediction result and the corresponding second measured internal resistance data to obtain the training set, including: According to the first error and each of the second errors, an importance parameter of each of the important feature data is determined, wherein the importance parameter represents the strength of the correlation between each of the important feature data and the preset initial model: In the formula, PI i represents the importance parameter corresponding to the i-th reference feature, E i represents the second error corresponding to the i-th reference feature, and E represents the first error; When the importance parameter is greater than a preset threshold, the important feature data corresponding to the current importance parameter is deleted from all the important feature data, and the set of the remaining important feature data in the important feature data is used as the training set.

4. The method according to claim 1, characterized in that: The method for obtaining the measured internal resistance includes: Obtaining a total voltage difference and a total current difference of each of the initial sample batteries at two adjacent moments at a preset SOC stage after the capacity division stage; According to the total voltage difference and the total current difference, the measured internal resistance corresponding to each of the initial sample batteries is calculated by the following formula: Where R represents the measured internal resistance, U K1 Indicates the total voltage difference at the previous moment, U K2 Represents the total voltage difference at the next moment, I K1 Represents the total current difference at the previous moment, I K2 Indicates the total current difference at the next moment.

5. The method according to claim 4, characterized in that The method further comprises: In the process of predicting the internal resistance of the battery to be tested by using the internal resistance prediction model, some of the batteries to be tested are extracted as calibration sample batteries, and the internal resistance of the calibration sample batteries is measured to obtain the calibrated measured internal resistance. When the error between the calibrated measured internal resistance and the internal resistance of the battery to be tested predicted by the internal resistance prediction model is greater than a preset error, a training set obtained based on the open-circuit potential curve data of the positive and negative electrodes of the calibration sample batteries and the production line data is used as a new training set, and the calibrated measured internal resistance is used as the corresponding new label value. The internal resistance prediction model is trained using the new training set and the new label value to update the internal resistance prediction model.

6. The method according to claim 5, characterized in that Before screening the influence of each of the initial feature data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set, the method further includes: Preprocessing each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data to obtain the corresponding preprocessed initial characteristic data and the positive and negative electrode open circuit potential curve data; Then, the influence of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance is screened to obtain a training set including: According to the influence of the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data on the internal resistance of the battery, the pre-processed initial characteristic data and the positive and negative electrode open circuit potential curve data are screened to obtain the training set.

7. The method according to claim 1, characterized in that The initial characteristic data includes at least one of the following parameters: The preset process parameters in the battery production process, the insulation withstand voltage test curve parameters of the uninjected battery, the dQdv curve parameters calculated according to the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process, the segmentation mechanism parameters after segmenting the voltage curve of the formation process and the voltage curve of the charging section of the capacity division process based on the dQdv curve, and the dQdt curve parameters calculated based on the battery temperature curve of the formation process and the battery temperature curve of the charging section of the capacity division process.

8. The method according to claim 7, characterized in that: The preset process parameters include at least one of the slurry viscosity, coating and drying temperature, rolling thickness, pole piece turnover time and injection volume in the battery production process.

9. A battery internal resistance prediction device, characterized in that: The device comprises: A first acquisition unit is used to acquire the positive and negative electrode open circuit potential curve data and production line data of the initial sample battery, wherein the production line data represents the parameters that affect the battery internal resistance during the production and internal resistance detection stages of the initial sample battery; A feature extraction unit, configured to extract features from the production line data to obtain initial feature data corresponding to each item of the production line data of each of the initial sample batteries; A screening unit, used to screen the influence of each item of the initial characteristic data and the positive and negative electrode open circuit potential curve data on the battery internal resistance to obtain a training set; A label value extraction unit, which uses the measured internal resistance of each of the initial sample batteries after the capacity division stage as a label value; A training unit trains the preset initial model using the training set and the corresponding flag value to obtain an internal resistance prediction model for battery internal resistance prediction, so as to use the internal resistance prediction model to predict the internal resistance of the battery to be tested.

10. An electronic device, characterized in that: The electronic device comprises a processor and a memory coupled to each other, wherein the memory stores a computer program. When the computer program is executed by the processor, the electronic device executes the method according to any one of claims 1 to 8.

Images