Battery health state diagnosis method and device, electronic equipment and storage medium

Abstract

The invention provides a battery health state diagnosis method and device, electronic equipment and a storage medium, and relates to the technical field of battery detection, and the method comprises the steps: obtaining battery data, and carrying out the data preprocessing, and obtaining a standard data set; constructing a state-of-charge-open-circuit voltage curve of the battery cell; according to the standard data set and the charge state-open-circuit voltage curve, estimating the single-point health degree of each battery cell in the single charging segment; obtaining a health degree sequence according to the single-point health degree, and performing data enhancement on the health degree sequence; performing double-layer filtering on the health degree sequence after data enhancement to obtain an optimal estimation health degree sequence; and diagnosing the health state of the battery according to the optimal estimated health degree sequence. According to the method, a self-adaptive updating mode of the health state degree is realized, health stable estimation of the whole life cycle under the actual working condition can be generalized, batteries of different battery types can be generalized, and the accuracy, stability and generalization of the model are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery detection, and in particular to a battery health status diagnosis method, device, electronic equipment and storage medium. Background Art

[0002] With the rapid development of new energy vehicles, lithium-ion batteries have been widely used due to their outstanding advantages such as high energy density, low self-discharge rate, long cycle life, energy saving and environmental protection. With the calendar and recycling of lithium-ion batteries, lithium-ion batteries are prone to problems such as reduced capacity, reduced performance, and deterioration of safety and stability. It is of great engineering significance to carry out health assessment and health diagnosis research on lithium-ion power batteries. However, once a lithium-ion battery is installed in a car, the maximum available capacity of the battery cannot be directly measured. At the same time, there are many complex chemical reactions and physical changes inside the lithium-ion battery, making it more complicated to accurately quantify the health status.

[0003] Under actual working conditions, existing methods mainly collect real-time information such as battery voltage, current and temperature through the battery management system on the vehicle side for prediction. By establishing an equivalent circuit model or life prediction model of the battery, the battery health attenuation trend at different rates, temperatures and discharge depths is predicted, thereby estimating the battery health in different cycles. However, this method is highly dependent on the battery model. As the battery ages, the battery model parameters change fundamentally, and complex electrochemical changes cannot be accurately represented by the battery model. In particular, adaptive model updates and iterations cannot be performed in the middle and late stages, and the accuracy and stability of the prediction method are greatly reduced. The battery aging model trained under laboratory conditions cannot be generalized to real conditions, and its prediction accuracy and stability cannot achieve the expected results over the entire life cycle. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a battery health status diagnosis method, device, electronic device and storage medium, which effectively solve the problems of low accuracy and stability of existing prediction methods.

[0005] In a first aspect, the present invention provides a battery health status diagnosis method, the method comprising:

[0006] Obtain battery data for data preprocessing to obtain a standard data set;

[0007] Construct the state of charge-open circuit voltage curve of the battery cell;

[0008] Estimate the single-point health of each battery cell in a single charging segment according to the standard data set and the state of charge-open circuit voltage curve;

[0009] Obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence;

[0010] Performing double-layer filtering on the health sequence after data enhancement to obtain an optimal estimated health sequence;

[0011] The health state of the battery is diagnosed based on the optimal estimated health state sequence.

[0012] Furthermore, after obtaining the optimal estimated health, the method further includes:

[0013] The state of charge-open circuit voltage curve is parameter-reconstructed according to the optimal estimated health sequence to obtain the state of charge-open circuit voltage curve under the optimal health state.

[0014] Furthermore, estimating the single-point health of each battery cell in a single charging segment according to the state of charge-open circuit voltage curve includes:

[0015] Acquire a charging segment in which there is a static condition at the beginning and end of charging in the charging condition;

[0016] Inputting the first voltage of each battery cell at the start of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the first state of charge of each battery cell;

[0017] Inputting the second voltage of each battery cell at the end of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the second state of charge of each battery cell;

[0018] Accumulating the ampere-hour integration of the current of the charging segment to obtain the real-time charging capacity of the charging segment;

[0019] The single-point health of each battery cell is calculated according to the first state of charge, the second state of charge and the real-time charging capacity.

[0020] Furthermore, obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence, includes:

[0021] Build a variational autoencoder model;

[0022] Perform time series calculation according to the single point health to obtain a time series health matrix;

[0023] Training the variational autoencoder model according to the time series health matrix to obtain an optimal variational autoencoder model;

[0024] The health sequence is constructed according to the single-point health, and the health sequence is input into the optimal variational autoencoder model for data enhancement.

[0025] Furthermore, the health sequence after data enhancement is subjected to double-layer filtering to obtain an optimal estimated health sequence, including:

[0026] Construct a Gaussian process regression model;

[0027] Using the Gaussian process regression model, filtering the global sequence points of the health sequence after data enhancement to obtain a sequence estimation value;

[0028] Eliminate noise points in the health degree sequence according to the sequence estimation value to obtain a non-noise point sequence;

[0029] The Gaussian process regression model is used to perform local filtering on the non-noise point sequence to obtain the optimal estimated health degree sequence.

[0030] Further, diagnosing the health status of the battery according to the optimal estimated health status sequence includes:

[0031] Calculating the decay rate of each battery cell according to the optimal estimated health sequence;

[0032] Calculate the health outlier degree of each battery cell according to the optimal estimated health sequence;

[0033] The health state of the battery is diagnosed according to the decay rate and / or the health state outlier degree.

[0034] Further, the parameter reconstruction of the state of charge-open circuit voltage curve according to the optimal estimated health sequence to obtain the state of charge-open circuit voltage curve in the optimal health state includes:

[0035] Calibrate the state of charge of each battery cell according to the optimal estimated health sequence;

[0036] Counting the open circuit voltage value of each battery cell according to the calibrated state of charge and the state of charge-open circuit voltage curve to obtain a state of charge-open circuit voltage matrix;

[0037] The state of charge-open circuit voltage matrix is ​​aggregated to obtain a state of charge-open circuit voltage curve in an optimal health state.

[0038] In a second aspect, the present invention provides a battery health status diagnosis device, the device comprising:

[0039] A data acquisition module is used to acquire battery data for data preprocessing to obtain a standard data set;

[0040] A curve construction module, used to construct a state of charge-open circuit voltage curve of a battery cell;

[0041] A data estimation module, used for estimating the single-point health of each battery cell in a single charging segment according to the standard data set and the state of charge-open circuit voltage curve;

[0042] A data enhancement module, used for obtaining a health sequence according to the single point health, and performing data enhancement on the health sequence;

[0043] A data filtering module, used for performing double-layer filtering on the health sequence after data enhancement to obtain an optimal estimated health sequence;

[0044] A state diagnosis module is used to diagnose the health state of the battery according to the optimal estimated health state sequence.

[0045] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the battery health status diagnosis method as described in the first aspect of the present invention.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the battery health status diagnosis method as described in the first aspect of the present invention.

[0047] The battery health status diagnosis method, device, electronic device and storage medium provided by the present invention adopt a method of estimating the optimal health status to complete the dynamic reconstruction of the state of charge-open circuit voltage curve, realize the self-update of model parameters, and at the same time, through the data enhancement method for sparse data under realistic working conditions, significantly improve the generalization effect and stability of the model under realistic working conditions. Based on the double-layer filtering method, an adaptive update method of the health status is realized, which can generalize the healthy and stable estimation of the whole life cycle under realistic working conditions, and can generalize batteries of different battery types, effectively improving the accuracy, stability and generalization of the model. The health diagnosis of the battery cells or modules in the battery pack can be realized, and the health status diagnosis at the cell level also provides strong support for the maintenance and abnormal diagnosis of the battery cells or modules. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 This is a first schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0050] Figure 2 This is a second schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0051] Figure 3 This is a third schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0052] Figure 4 Schematic diagram of the architecture of a variational autoencoder model in an embodiment of the present invention;

[0053] Figure 5 is the original health sequence in the embodiment of the present invention;

[0054] Figure 6 is a new health sequence generated by the variational autoencoder model in an embodiment of the present invention;

[0055] Figure 7 is a fourth schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0056] Figure 8 is a schematic diagram of the result after data filtering in an embodiment of the present invention;

[0057] Fig. 9 is a fifth schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0058] Fig.10 is a sixth schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0059] Fig.11 This is the seventh schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention;

[0060] Fig.12 is a schematic diagram of reconstructing a state of charge-open circuit voltage curve in an embodiment of the present invention;

[0061] Fig.13 is a schematic diagram of the results of the battery health status diagnosis device provided by an embodiment of the present invention;

[0062] Fig.14 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention.

[0063] Description of main component symbols:

[0064] 800, battery health status diagnosis device; 810, data acquisition module; 820, curve construction module; 830, data estimation module; 840, data enhancement module; 850, data filtering module; 860, status diagnosis module; 900, electronic device; 910, processor; 920, communication interface; 930, memory; 940, communication bus. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be noted that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0066] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0068] Battery health (State Of Health, SOH) is used to evaluate the maximum intrinsic capacity that can be released by the battery in its current state. The vehicle-side SOH prediction is mainly carried out by collecting real-time information such as battery voltage, current and temperature through the battery management system (BMS). By establishing an equivalent circuit model or life prediction model for the battery, the SOH decay trend of the battery at different rates, temperatures, and discharge depths is predicted, thereby estimating the battery SOH in different cycles. However, this method is highly dependent on the battery model. As the battery ages, the battery model parameters change essentially, and complex electrochemical changes cannot be accurately represented by the battery model. In particular, adaptive model updates and iterations cannot be performed in the middle and late stages, and the accuracy and stability of this prediction method are greatly reduced. At the same time, due to the limited computing resources of the local embedded system, the complexity of the battery model is greatly reduced, and it is impossible to accurately predict the full life cycle of different batteries, so it is impossible to accurately diagnose and evaluate the battery.

[0069] Therefore, relevant technicians upload the vehicle-side battery data to the cloud, and the cloud performs analysis and SOH prediction based on the battery operation data. The existing method is mainly based on the aging data of laboratory cells, and establishes a mapping relationship between the cell and aging factors such as cycles and calendars, so as to predict the battery's SOH based on the aging factors. However, this is a prediction of an ideal working condition. The laboratory working condition (constant current charging, and full charge and discharge state) is extremely simple and cannot be generalized to the SOH estimation under actual working conditions. Under actual working conditions, the battery's current, temperature, and voltage are extremely variable, and the battery operation data is relatively sparse. Because the vehicle side may be running or may be stagnant for a long time, the sampling frequency of the actual vehicle data is greatly reduced, and the sampling cycle is long, which cannot be compared with the stable and single working conditions in the laboratory. Therefore, it is impossible to extract the same aging factors as in the laboratory under actual working conditions, so that the battery aging model trained under laboratory working conditions cannot be generalized to actual working conditions, and its prediction accuracy and stability cannot achieve the expected results over the entire life cycle.

[0070] Example 1

[0071] The embodiment of the present invention provides a battery health status diagnosis method, which effectively solves the problem that the accuracy and stability of existing prediction methods cannot meet the full life cycle prediction requirements of the battery. Figure 1 FIG. 1 is a first schematic diagram of a battery health status diagnosis method process according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0072] S100: Obtain battery data, perform data preprocessing, and obtain a standard data set.

[0073] In an embodiment of the present invention, the health of the vehicle's lithium-ion battery is predicted based on a cloud scenario. Under real working conditions, the vehicle's power or energy storage system will continuously generate new battery data during operation. The BMS collects, monitors, and manages the battery's voltage, current, and temperature, and uploads the data to the cloud system in real time. The cloud system and the vehicle side comply with the GB32960 protocol (a national standard protocol for transmitting data from a local end to the cloud), and the cloud system acquires and stores battery data in real time for big data analysis.

[0074] After the cloud obtains the battery data, it mainly analyzes the cell voltage data, temperature data, current data and charge and discharge status, and cleans the data of abnormal values. In an embodiment of the present invention, the cell voltage data can be a voltage value generated by each cell at each moment. The series system composed of multiple cells forms a voltage list with all voltage data. Similarly, a temperature list, a current list and a charge and discharge status list can be formed. Abnormal values ​​include but are not limited to illegal values, jump values, sampling abnormal values, mutation values, signal jitter and other abnormal type values. For illegal values ​​(for example, the reasonable range of the cell voltage is within 1.8V~4.5V, and the value beyond the range is regarded as an illegal value) are directly eliminated. For jump values, mutation values ​​and signal jitter values, forward filling, average filling or weighted filling are performed. For sampling abnormal values, the event range is eliminated.

[0075] After data cleaning, the operating conditions are classified according to the vehicle status and charging status, mainly divided into charging condition, static condition and discharging condition. Specifically, if the battery is in a charging state, it is marked as a charging condition; if the battery is in an uncharged state and a discharging state, it is marked as a discharging condition; if the battery is in a uncharged and undischarged state, it is marked as a static condition.

[0076] S200: construct a state of charge-open circuit voltage curve of the battery cell.

[0077] When a lithium-ion battery is in an open circuit state, that is, no current passes through the external circuit, the chemical reaction inside the battery will tend to a balanced state, and the potential difference between the positive and negative electrodes reaches a stable value, which is the open circuit voltage (OCV). There is a certain mapping relationship between the state of charge (SOC) and OCV of lithium-ion batteries, which is the state of charge-open circuit voltage curve (SOC-OCV curve). This curve is used to verify the current SOC of the battery cell. Generally, the OCV of a lithium battery increases with the increase of SOC, showing a specific curve relationship. The real SOC of the battery cell can be obtained through the OCV collected by the BMS.

[0078] In the embodiment of the present invention, the SOC and OCV of the initial battery cell may be measured in a laboratory environment, and then the SOC-OCV curve may be calibrated according to the measured SOC and OCV to obtain an initial SOC-OCV curve.

[0079] S300, estimating the single-point health of each battery cell in a single charging segment according to a standard data set and a state of charge-open circuit voltage curve.

[0080] In the embodiment of the present invention, a charging condition under a real working condition is selected, and a single SOH estimation is performed on a charging segment under a static condition at the beginning and the end of charging. Figure 2 FIG. 2 is a second schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention. Figure 2 As shown, the calculation of single point health includes the following steps:

[0081] S310: Acquire a charging segment in which a static condition exists at the beginning and the end of charging in a charging condition.

[0082] In an embodiment of the present invention, a charging segment is selected in which a static condition duration exceeds a first time threshold at the start of charging and a static condition duration exceeds a second time threshold at the end of charging. The first time threshold and the second time threshold can be set according to actual conditions.

[0083] S320 , inputting the first voltage of each battery cell at the start of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the first state of charge of each battery cell.

[0084] When there is a static condition at the start of charging in the charging condition and the duration exceeds the first time threshold, the first voltage of each battery cell at the start of charging in the charging segment is obtained to form a first voltage list, and the first voltage list is input into the SOC-OCV curve to obtain the first state of charge of each battery cell to form a first state of charge list.

[0085] S330, inputting the second voltage of each battery cell at the end of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the second state of charge of each battery cell.

[0086] When the static working condition exists for a period exceeding the second time threshold at the end of charging, the second voltage of each battery cell at the end of charging in the charging segment is obtained to form a second voltage list, and the second voltage list is input into the SOC-OCV curve to obtain the second state of charge of each battery cell to form a second state of charge list.

[0087] S340, accumulating the ampere-hour integral of the current of the charging segment to obtain the real-time charging capacity of the charging segment.

[0088] The ampere-hour integration method integrates the charging current of the charging segment. The real-time charging capacity of the charging segment can be obtained by integrating the current and time. The basic principle is that the change in battery charge is proportional to the integral of the current passing through the battery.

[0089] S350, calculating and obtaining the single-point health of each battery cell according to the first state of charge, the second state of charge and the real-time charging capacity.

[0090] In the embodiment of the present invention, the single-point health calculation formula of each battery cell is as follows:

[0091]

[0092] In the above formula, SOH 1 Represents the health of a single point, C E Indicates real-time charging capacity, SOC 1 Indicates the first state of charge, SOC 2 Indicates the second state of charge, C N Indicates the rated capacity of the battery when it leaves the factory.

[0093] The single-point health of each battery cell is calculated to form a single-point health matrix. For example, if a battery includes 240 batteries, the single-point health matrix has 240 single-point healths.

[0094] S400: Obtain a health sequence according to the single-point health, and perform data enhancement on the health sequence.

[0095] Certain trigger conditions are required to meet the single-point health estimation, that is, there is a static condition before and after the charging segment and the charging depth exceeds the threshold, where the charging depth is SOC 2 -SOC 1 In real working conditions, it is impossible to have calculations that meet the trigger conditions every day. For example, there may only be 10 estimates triggered in a month, resulting in a large amount of sparse data. Therefore, it is necessary to capture missing values ​​based on a large number of single-point health trends and automatically generate missing points for subsequent calculation of the optimal estimated battery health.

[0096] Figure 3 FIG. 3 is a third schematic diagram of the battery health status diagnosis method process provided by an embodiment of the present invention. Figure 3 As shown, data enhancement specifically includes the following steps:

[0097] S410. Construct a variational autoencoder model.

[0098] In an embodiment of the present invention, a generative neural network model variational autoencoder model is adopted. Figure 4 Schematic diagram of the architecture of the variational autoencoder model in the embodiment of the present invention. Figure 4 As shown in Figure 1, the architecture of the variational autoencoder model mainly includes an encoder and a decoder. The encoder is responsible for compressing the input data into a latent space representation, and the decoder reconstructs the original data from this representation.

[0099] The loss function of the variational autoencoder model consists of two parts, namely the reconstruction loss Loss_rec and the KL divergence loss Loss_KL, where:

[0100] The purpose of reconstruction loss is to make the reconstructed data as close as possible to the original input data, so as to ensure that the model can effectively learn the characteristics and distribution of the data, so that the data can be accurately restored during decoding. In the embodiment of the present invention, mean square error MSE is used for characterization, and the functional expression of reconstruction loss is as follows:

[0101]

[0102] In the above formula, Loss rec Represents the function of reconstruction loss, n represents the number of samples, x i represents the time series window data of the i-th input, Represents the average value of the input time series window data.

[0103] KL divergence loss is used to measure the difference between two probability distributions, that is, the difference between the distribution of the latent variable Z and the prior distribution N(0,1). The role of KL divergence loss is to make the distribution of the latent variable Z output by the encoder as close to the prior distribution as possible, thereby ensuring the continuity and standardization of the latent space, making sampling and generating new data in the latent space more reasonable and effective, and effectively preventing overfitting. The functional expression of KL divergence loss is as follows:

[0104]

[0105] In the above formula, Loss KL Represents the KL divergence loss function, n represents the number of samples, μ i represents the mean value of the i-th sample, σ i represents the standard deviation of the ith sample.

[0106] The total loss function of the model is the weighted sum of the reconstruction loss and the KL divergence loss. The expression of the total loss function of the model is as follows:

[0107] Loss model =Loss rec +α·Loss EL

[0108] In the above formula, Loss model Represents the total loss function of the model, α represents a hyperparameter used to measure Loss rec and Loss KL importance.

[0109] By minimizing this total loss function, the variational autoencoder model can simultaneously generate samples similar to the original data and learn the distribution of the latent space.

[0110] S420: Perform time series calculation according to the single point health to obtain a time series health matrix.

[0111] The time series health matrix is ​​obtained by performing time series calculation on the health of a single point within a certain time period, and the time period can be one month or several months.

[0112] S430: training a variational autoencoder model according to the time series health matrix to obtain an optimal variational autoencoder model.

[0113] The time series health matrix is ​​converted into a time series data curve, and a large number of time series data curves are input into the fully connected neural network to encode the input features. The neural network is used to learn the potential space distribution of the input, and the input features are resampled to obtain the latent variable Z. The resampled latent variable Z is then decoded, and the loss function is used to minimize the distribution of output and input, and finally the optimal variational autoencoder model is obtained. No additional labels are required during the training process of the variational autoencoder model, so there is no additional label acquisition cost.

[0114] S440, constructing a health sequence according to the single-point health, and inputting the health sequence into the optimal variational autoencoder model for data enhancement.

[0115] Obtain the single-point health under real working conditions and construct a health sequence, which is usually a sequence with data missing points. Input the health sequence into the optimal variational autoencoder model to estimate the data generation probability near the data missing point, and regenerate a new health sequence based on the long sequence rules learned by the optimal variational autoencoder model.

[0116] Figure 5 is the original health sequence in the embodiment of the present invention, Figure 6 is a new health sequence generated by the variational autoencoder model in the embodiment of the present invention, according to Figure 5 and Figure 6 Analysis shows that the new health sequence has more data and a better SOH decay trend, which greatly improves the generalization performance and stability of the SOH optimal estimation.

[0117] S500, performing double-layer filtering on the health sequence after data enhancement to obtain the optimal estimated health sequence.

[0118] like Figure 6 As shown in FIG. 1 , the health sequence after data enhancement is affected by different working conditions and data sampling periods, and exhibits different degrees of jitter and noise. Therefore, it is necessary to filter the new health sequence in order to obtain the optimal estimated health sequence. Figure 7 4 is a schematic diagram of a battery health status diagnosis method according to an embodiment of the present invention. Figure 7 As shown, data filtering specifically includes the following steps:

[0119] S510. Construct a Gaussian process regression model.

[0120] Gaussian Process Regression (GPR) is a nonparametric regression method based on Bayesian statistics. It can use Gaussian process to model the potential distribution of data and give predicted values ​​and corresponding uncertainty estimates. Gaussian process is a random process in which any finite number of random variables obeys a joint Gaussian distribution. A Gaussian process is completely defined by the mean function and the kernel function, that is:

[0121] f(x)~GP(m(x),k(x,x'))

[0122] In the above formula, f(x) represents the optimal estimated health, m(x) represents the mean function, and k(x,x') represents the kernel function, which is used to describe the similarity between input sequences.

[0123] In the embodiment of the present invention, according to the timing SOH matrix of a given window Find the optimal estimated health, where x i and i Represents the corresponding time series data, and the window threshold can be set according to the actual situation. The kernel function uses the Markov kernel function, and its expression is as follows:

[0124]

[0125] In the above formula, represents the signal variance, which is used to control the overall scaling of the function, and l represents the length scale parameter.

[0126] Specifically, at each time step x, the known time series SOH matrix As training data, train the Gaussian process regression model, and then t Make a prediction to get μ t and where μ t is the predicted value at this time step, It is the uncertainty estimate of the prediction result. A smaller uncertainty estimate means that the model has a higher confidence in the current prediction, otherwise the uncertainty is larger.

[0127] S520: Filter the global sequence points of the health sequence after data enhancement using a Gaussian process regression model to obtain a sequence estimation value.

[0128] In the embodiment of the present invention, a double-layer Gaussian process regression model is used for filtering, and the first layer of the trained Gaussian process regression model is used to predict the global sequence points in the health sequence after data enhancement to obtain the sequence estimation value r 1 .

[0129] S530: Remove noise points in the health degree sequence according to the sequence estimation value to obtain a non-noise point sequence.

[0130] Since there are noise points in the health sequence after data enhancement, when the health sequence after data enhancement exists |r i -r 1 |≥r 0 , then the noise point is actively removed, where r i Represents the single point health of the noise point, r 1 represents the estimated value of the sequence, r 0 Represents the estimated value threshold, which is set according to the actual situation. After repeating the entire sequence, a non-noise point sequence can be obtained.

[0131] S540. Use a Gaussian process regression model to perform local filtering on the non-noise point sequence to obtain an optimal estimated health degree sequence.

[0132] The second layer of the Gaussian process regression model is used to perform local filtering on the non-noise point sequence to obtain the optimal estimated health sequence excluding noise points.

[0133] Figure 8 is a schematic diagram of the result after data filtering in an embodiment of the present invention, such as Figure 8 As shown in Figure 2, the health sequence after data enhancement consists of single-point health and data enhancement points. The result after double-layer filtering is the optimal estimated health sequence curve. Figure 8 It can be seen that the confidence of this filtering method is high and it is not affected by local noise points. The prediction stability and accuracy have reached a good level, and the generalization performance has been greatly improved. At the same time, the attenuation trend of SOH varies under different working conditions, and the attenuation trend does not decrease linearly with the cycle or calendar, and there is a strong nonlinear relationship. This also shows that this filtering method has good accuracy and stability. With the continuous use of the battery, this method can be applied to the SOH estimation of the entire life cycle, and can be automatically updated and iterated.

[0134] S600: Diagnose the health status of the battery according to the optimal estimated health status sequence.

[0135] In an embodiment of the present invention, the optimal estimated health sequences of all battery cells are combined into a battery cell health list soh_list, and the SOH decay trend or outlier characteristics of each battery cell are tracked to diagnose the health status of the battery and issue early warnings for abnormally decayed batteries. Fig. 9 FIG. 5 is a fifth schematic diagram of a battery health status diagnosis method process according to an embodiment of the present invention. Fig. 9 As shown, the prediction of the battery health status includes the following steps:

[0136] S610: Calculate the decay rate of each battery cell according to the optimal estimated health sequence.

[0137] In the embodiment of the present invention, in units of time periods, soh_list_1 represents a list of battery cell health in a first time period, soh_list_2 represents a list of battery cell health in a second time period, and the calculation formula of the decay rate is as follows:

[0138] soh_rate_list=soh_list_1-soh_list_2

[0139] In the above formula, soh_rate_list represents the monthly attenuation rate list of each battery cell.

[0140] S620: Calculate the health outlier degree of each battery cell according to the optimal estimated health sequence.

[0141] In the embodiment of the present invention, the health outlier degree of each battery cell is calculated according to the battery cell health list soh_list, and the calculation formula is as follows:

[0142] soh_list_disp=(soh_list-soh_list_avg) / soh_list_std

[0143] In the above formula, soh_list_disp represents the health outlier list, soh_list_avg represents the average value of the battery health list, and soh_list_std represents the standard deviation of the battery health list.

[0144] S630: Diagnose the health status of the battery according to the decay rate and / or the health status outlier degree.

[0145] Optionally, if the decay rate soh_rate_list_i of the ith battery cell exceeds the decay rate threshold, and / or if soh_list_max-soh_list_min exceeds the outlier threshold and the health outlier soh_list_disp_i of the ith battery cell exceeds the outlier threshold, a battery health status alarm is issued and the alarm information is pushed to the client to locate and identify the specific abnormal battery cell for offline repair or update.

[0146] As a preferred embodiment of the present invention, Fig.10 6 is a schematic diagram of a battery health status diagnosis method according to an embodiment of the present invention. Fig.10 As shown, the method also includes the following steps:

[0147] S700 , reconstructing parameters of a state of charge-open circuit voltage curve according to an optimal estimated health sequence to obtain a state of charge-open circuit voltage curve in an optimal health state.

[0148] As the battery ages, the SOC-OCV curve will differ significantly. The initial SOC-OCV curve obtained in the laboratory is completely unable to meet the needs of SOH estimation. Therefore, the SOC-OCV curves of all battery cells need to be reconstructed and verified to meet the SOH estimation of the entire life cycle. Fig.11 FIG. 7 is a seventh schematic diagram of a battery health status diagnosis method process according to an embodiment of the present invention. Fig.11 As shown, parameter reconstruction specifically includes the following steps:

[0149] S710: Calibrate the state of charge of each battery cell according to the optimal estimated health sequence.

[0150] Since there is a large error between the SOC sent by the BMS and the actual SOC, it is necessary to calibrate the SOC based on the ampere-hour integral under the charge and discharge conditions based on the optimal estimated health sequence. Specifically, the charge state of discharge is calibrated according to the following formula:

[0151] SOC F =1-C F / (SOH i ×C N )

[0152] In the above formula, SOC F Indicates the state of charge during discharge, C F Indicates the discharge capacity, which is obtained by integrating the current by ampere-hour. SOH i represents the optimal estimated health of the ith battery cell, C N Indicates the rated capacity of the battery when it leaves the factory.

[0153] The charge state of the battery is calibrated according to the following formula:

[0154] SOC C =SOC 0 +C C / (SOH i ×C N )

[0155] In the above formula, SOC C Indicates the state of charge, SOC 0 Indicates the initial value of the state of charge at the last moment of discharge, C C Indicates the charging capacity, which is obtained by integrating the ampere-hours of the current.

[0156] S720 , counting the open circuit voltage value of each battery cell according to the calibrated state of charge and the state of charge-open circuit voltage curve to obtain a state of charge-open circuit voltage matrix.

[0157] According to the state of charge of all calibrated battery cells, the open circuit voltage value of each battery cell is counted through the state of charge-open circuit voltage curve, and the state of charge-open circuit voltage matrix in the big data scenario is obtained by statistics.

[0158] S730 , aggregate the state of charge-open circuit voltage matrix to obtain a state of charge-open circuit voltage curve in an optimal health state.

[0159] The state of charge-open circuit voltage matrix is ​​aggregated with big data, the state of charge-open circuit voltage curve is re-acquired and corrected, and the state of charge-open circuit voltage curve under the optimal health state is obtained. Fig.12 is a schematic diagram of the reconstruction of the state of charge-open circuit voltage curve in an embodiment of the present invention, such as Fig.12 As shown, the closed-loop update iteration is performed at every 5% attenuation. As the SOH decreases from a high level, the curve shifts significantly, which will have a significant impact on the SOH estimation. The initial curve is completely unable to meet the subsequent SOH estimation.

[0160] The battery health status diagnosis method provided in this embodiment is used to verify the battery samples. The root mean square error RMSE is ≤2% when verified by a large number of battery samples under realistic operating conditions, which is more than 30% higher than the existing methods. At the same time, it can generalize batteries of different battery types. This method uses a method of estimating the optimal health state to complete the dynamic reconstruction of the state of charge-open circuit voltage curve, realizes the self-update of model parameters, and at the same time, through the data enhancement method for sparse data under realistic working conditions, significantly improves the generalization effect and stability of the model under realistic working conditions. Based on the double-layer filtering method, an adaptive update method of the health state is realized, which can generalize the health and stability estimation of the entire life cycle under realistic working conditions.

[0161] Example 2

[0162] Based on the same technical concept, an embodiment of the present invention provides a battery health status diagnosis device. Fig.13 is a schematic diagram of the results of the battery health status diagnosis device provided by an embodiment of the present invention, such as Fig.13 As shown, the battery health status diagnosis device 800 includes:

[0163] The data acquisition module 810 is used to acquire battery data for data preprocessing to obtain a standard data set;

[0164] A curve construction module 820, used to construct a state of charge-open circuit voltage curve of a battery cell;

[0165] A data estimation module 830, for estimating a single-point health of each battery cell in a single charging segment according to a standard data set and a state of charge-open circuit voltage curve;

[0166] A data enhancement module 840 is used to obtain a health sequence according to the single-point health, and perform data enhancement on the health sequence;

[0167] The data filtering module 850 is used to perform double-layer filtering on the health sequence after data enhancement to obtain the optimal estimated health sequence;

[0168] The state diagnosis module 860 is used to diagnose the health state of the battery according to the optimal estimated health state sequence.

[0169] The battery health status diagnosis device provided by the embodiment of the present invention can generalize the health stability estimation of the whole life cycle under realistic working conditions, and can also generalize batteries of different battery types, effectively improving the accuracy, stability and generalization of the model. The health diagnosis of the battery cells or modules in the battery pack can be realized, and the health status diagnosis of the battery cell level also provides strong support for the maintenance and abnormal diagnosis of the battery cells or modules.

[0170] It can be understood that the implementation method of the battery health status diagnosis method described in the above embodiment 1 is also applicable to this embodiment and can achieve the same technical effect, so it will not be repeated here.

[0171] Example 3

[0172] Based on the same concept, an embodiment of the present invention further provides an electronic device, Fig.14 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Fig.14 As shown, the electronic device 900 may include: a processor 910, a communication interface 920, a memory 930 and a communication bus 940, wherein the processor 910, the communication interface 920 and the memory 930 communicate with each other through the communication bus 940. The processor 910 may call the logic instructions in the memory 930 to execute the steps of the battery health status diagnosis method described in the above embodiments. For example, it includes:

[0173] S100, obtaining battery data for data preprocessing to obtain a standard data set;

[0174] S200, constructing a state of charge-open circuit voltage curve of the battery cell;

[0175] S300, estimating the single-point health of each battery cell in a single charging segment according to a standard data set and a state of charge-open circuit voltage curve;

[0176] S400, obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence;

[0177] S500, performing double-layer filtering on the health sequence after data enhancement to obtain the optimal estimated health sequence;

[0178] S600: Diagnose the health status of the battery according to the optimal estimated health status sequence.

[0179] The processor 910 may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0180] In addition, the logic instructions in the above-mentioned memory 930 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0181] The memory 930 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0182] Example 4

[0183] Based on the same concept, an embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program, and the computer program includes at least one code segment, which can be executed by a main control device to control the main control device to implement the steps of the battery health status diagnosis method described in the above embodiments. For example, it includes:

[0184] S100, obtaining battery data for data preprocessing to obtain a standard data set;

[0185] S200, constructing a state of charge-open circuit voltage curve of the battery cell;

[0186] S300, estimating the single-point health of each battery cell in a single charging segment according to a standard data set and a state of charge-open circuit voltage curve;

[0187] S400, obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence;

[0188] S500, performing double-layer filtering on the health sequence after data enhancement to obtain the optimal estimated health sequence;

[0189] S600: Diagnose the health status of the battery according to the optimal estimated health status sequence.

[0190] Based on the same technical concept, an embodiment of the present invention further provides a computer program, which is used to implement the above method embodiment when executed by a main control device.

[0191] The computer program may be stored in whole or in part on a computer-readable storage medium packaged with the processor, or may be stored in whole or in part on a memory not packaged with the processor.

[0192] Based on the same technical concept, an embodiment of the present invention further provides a processor, which is used to implement the above method embodiment. The above processor can be a chip.

[0193] In summary, the battery health status diagnosis method, device, electronic device and storage medium provided by the present invention adopt a method of estimating the optimal health status to complete the dynamic reconstruction of the state of charge-open circuit voltage curve, realize the self-update of model parameters, and at the same time, through the data enhancement method for sparse data under actual working conditions, significantly improve the generalization effect and stability of the model under actual working conditions. Based on the double-layer filtering method, an adaptive update method of the health status is realized, which can generalize the health and stability estimation of the whole life cycle under actual working conditions, and can generalize batteries of different battery types, effectively improving the accuracy, stability and generalization of the model. The health diagnosis of the battery cells or modules in the battery pack can be realized, and the health status diagnosis at the cell level also provides strong support for the maintenance and abnormal diagnosis of the battery cells or modules.

[0194] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0195] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A battery health status diagnosis method, characterized in that: The method comprises: Obtain battery data for data preprocessing to obtain a standard data set; Construct the state of charge-open circuit voltage curve of the battery cell; Estimate the single-point health of each battery cell in a single charging segment according to the standard data set and the state of charge-open circuit voltage curve; Obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence; Performing double-layer filtering on the health sequence after data enhancement to obtain an optimal estimated health sequence; The health state of the battery is diagnosed based on the optimal estimated health state sequence.

2. The battery health status diagnosis method according to claim 1, characterized in that: After obtaining the optimal estimated health, the method further includes: The state of charge-open circuit voltage curve is parameter-reconstructed according to the optimal estimated health sequence to obtain the state of charge-open circuit voltage curve under the optimal health state.

3. The battery health status diagnosis method according to claim 1, characterized in that: The estimating the single-point health of each battery cell in a single charging segment according to the state of charge-open circuit voltage curve includes: Acquire a charging segment in which there is a static condition at the beginning and end of charging in the charging condition; Inputting the first voltage of each battery cell at the start of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the first state of charge of each battery cell; Inputting the second voltage of each battery cell at the end of charging in the charging segment into the state of charge-open circuit voltage curve to obtain the second state of charge of each battery cell; Accumulating the ampere-hour integration of the current of the charging segment to obtain the real-time charging capacity of the charging segment; The single-point health of each battery cell is calculated according to the first state of charge, the second state of charge and the real-time charging capacity.

4. The battery health status diagnosis method according to claim 3, characterized in that: The step of obtaining a health sequence according to the single-point health, and performing data enhancement on the health sequence, comprises: Build a variational autoencoder model; Perform time series calculation according to the single point health to obtain a time series health matrix; Training the variational autoencoder model according to the time series health matrix to obtain an optimal variational autoencoder model; The health sequence is constructed according to the single-point health, and the health sequence is input into the optimal variational autoencoder model for data enhancement.

5. The battery health status diagnosis method according to claim 4, characterized in that: The step of performing double-layer filtering on the health sequence after data enhancement to obtain an optimal estimated health sequence includes: Construct a Gaussian process regression model; Using the Gaussian process regression model, filtering the global sequence points of the health sequence after data enhancement to obtain a sequence estimation value; Eliminate noise points in the health degree sequence according to the sequence estimation value to obtain a non-noise point sequence; The Gaussian process regression model is used to perform local filtering on the non-noise point sequence to obtain the optimal estimated health degree sequence.

6. The battery health status diagnosis method according to claim 1, characterized in that: The diagnosing the health status of the battery according to the optimal estimated health status sequence includes: Calculating the decay rate of each battery cell according to the optimal estimated health sequence; Calculate the health outlier degree of each battery cell according to the optimal estimated health sequence; The health state of the battery is diagnosed according to the decay rate and / or the health state outlier degree.

7. The battery health status diagnosis method according to claim 2, characterized in that: The step of reconstructing parameters of the state of charge-open circuit voltage curve according to the optimal estimated health sequence to obtain the state of charge-open circuit voltage curve in an optimal health state includes: Calibrate the state of charge of each battery cell according to the optimal estimated health sequence; Counting the open circuit voltage value of each battery cell according to the calibrated state of charge and the state of charge-open circuit voltage curve to obtain a state of charge-open circuit voltage matrix; The state of charge-open circuit voltage matrix is ​​aggregated to obtain a state of charge-open circuit voltage curve in an optimal health state.

8. A battery health status diagnosis device, characterized in that: The device comprises: A data acquisition module is used to acquire battery data for data preprocessing to obtain a standard data set; A curve construction module, used to construct a state of charge-open circuit voltage curve of a battery cell; A data estimation module, used for estimating the single-point health of each battery cell in a single charging segment according to the standard data set and the state of charge-open circuit voltage curve; A data enhancement module, used for obtaining a health sequence according to the single point health, and performing data enhancement on the health sequence; A data filtering module, used for performing double-layer filtering on the health sequence after data enhancement to obtain an optimal estimated health sequence; A state diagnosis module is used to diagnose the health state of the battery according to the optimal estimated health state sequence.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes the computer program to implement the battery health status diagnosis method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the battery health status diagnosis method according to any one of claims 1 to 7 is implemented.

Images