Method for Estimating the Health State of Power Batteries in Battery Swap Stations Based on Small Data Samples

Abstract

The present invention discloses a method for estimating the health state of power batteries in a battery swapping station based on small data samples. By obtaining the usage data of the power batteries in the battery swapping station, temperature characteristic values, capacity increment, and charging time characteristic values are analyzed, the DTV and IC curves are analyzed. By using BiLSTM to process the features extracted from the original time series data by CNN, an attention mechanism is applied after the LSTM layer to dynamically select relevant hidden states in the entire time series, and transfer learning is used to solve the problem of limited data in the target domain. Bayesian optimization is used to optimize the model hyperparameters in the pre-training stage with sufficient data; the discharge characteristic values are obtained by analyzing the power battery data, the temperature characteristic values, capacity increment, and charging time characteristic values of the power battery are calculated, the proportion of abnormal data of the power battery is analyzed, the charge and discharge state value is obtained by analyzing the charge and discharge conditions of the power battery, and the health characteristic value is obtained by comprehensively analyzing the data.

Description

Technical Field

[0001] The present invention relates to the technical field of battery health assessment, and particularly to a method for estimating the health state of power batteries in battery swapping stations based on small data samples. Background Art

[0002] With the continuous progress of related technologies, the electrification of automobiles has shown a booming development trend. The electrification of automobiles has also become one of the key ways to contribute to the realization of carbon neutrality. However, the problem of slow charging still severely restricts the practicality of electric vehicles.

[0003] Adopting battery swapping technology is an effective solution to the problem of slow charging. However, for the power batteries used in vehicles with battery swapping technology, a battery often circulates among different battery swapping stations and vehicles, and its usage data is often very scattered, making it difficult to obtain relatively large and complete data for estimating the state of health (SOH) of the battery; the generalization of battery health state estimation methods is also one of the problems that need to be improved; the prediction accuracy of existing battery health state prediction methods highly depends on the underlying battery degradation model, and it is difficult to simultaneously consider both the model prediction accuracy and the generalization under different working conditions; data-driven methods are mechanism-free and non-parametric. Due to their powerful ability to handle non-linear mappings, they have higher flexibility in modeling than model-based methods; long short-term memory models, which are deep learning models with powerful data processing capabilities, have also been used for battery health state estimation. Usually, existing methods require more than 25% of the data over the entire battery life for parameter estimation and model training to achieve accurate capacity trajectory prediction. Machine learning usually assumes that the data in the training set and the test set have the same distribution. However, on-vehicle power batteries may experience various complex and changeable working conditions, including different temperatures, charge and discharge rates, environmental humidity, and frequent charge and discharge cycles. These working conditions will significantly affect the health state and performance of the battery, resulting in a decrease in prediction accuracy. Therefore, there is a lack of a prediction model that can adapt to various working conditions with a small amount of data support and has strong generalization ability.

[0004] Therefore, the present invention provides a method for estimating the health state of power batteries in battery swapping stations based on small data samples. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for estimating the health state of power batteries in battery swapping stations based on small data samples to solve at least one of the above-mentioned problems in the prior art.

[0006] In a first aspect, the present invention provides a method for estimating the health state of power batteries in battery swapping stations based on small data samples, including the following steps:

[0007] By obtaining the usage data of the power batteries in the battery swapping station, analyzing to obtain temperature characteristic values, capacity increment, and charging time characteristic values, and analyzing to obtain the DTV and IC curves;

[0008] By using the features extracted from the original time series data through processing CNN with BiLSTM, an attention mechanism is applied after the LSTM layer to dynamically select relevant hidden states throughout the time series;

[0009] Transfer learning is selected to solve the problem of limited data in the target domain. In the pre-training stage with sufficient data, Bayesian optimization is used to optimize the model hyperparameters for the development of the neural network model;

[0010] Discharge characteristic values are obtained through DTV and IC curve analysis, temperature characteristic values, capacity increment, and charging time characteristic values of the power battery are calculated, and the proportion of abnormal data of the power battery is analyzed;

[0011] The charge and discharge status values are obtained by analyzing the charge and discharge conditions of the power battery, and the health characteristic values are obtained through comprehensive data analysis for classification analysis.

[0012] Advantages of the present invention:

[0013] 1. A multi-feature analysis method combining DTV, IC, and CVT is used to more fully extract the features characterizing battery aging to obtain more accurate SOH prediction results. A network is constructed by combining multiple methods such as CNN-BiLSTM, attention mechanism, and Bayesian optimization to improve the learning efficiency and effect of the SOH prediction framework on battery data. The transfer learning method can achieve accurate prediction with only 10% of the data after sufficient pre-training, ensuring the generalization of the SOH prediction model, realizing the prediction using small sample data, being applicable to the situation of scattered and missing data of in-vehicle power batteries in battery swapping stations, and enhancing the flexibility and adaptability of the model.

[0014] 2. The temperature, voltage, and current of the power battery are monitored simultaneously to ensure the integrity and accuracy of the data. The overall charge and discharge performance score is obtained by integrating the discharge characteristic value FT and the charging time deviation ratio PC. Based on the abnormal data proportion YC and the charge and discharge status value CF, a comprehensive index for measuring battery health is obtained. According to whether the health characteristic value JK exceeds the preset threshold JK0, two different aging modes can be automatically distinguished for easy understanding and support the construction of a more accurate SOH prediction model for power batteries by machine learning algorithms. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of the steps of the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0017] Figure 2 It is a flowchart of the steps of the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 2 of the present invention;

[0018] Figure 3 It is a schematic structural diagram of a computer device for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 3 of the present invention;

[0019] Attached drawing notes: 3 - computer device, 301 - processor, 302 - memory, 303 - computer program;

[0020] Figure 4 It is a graph of the change of power battery data for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0021] Figure 5 It is a schematic diagram of transfer learning for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0022] Figure 6 It is a statistical chart of correlation coefficients for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0023] Figure 7 It is a prediction structure diagram for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0024] Figure 8 It is a statistical chart of lithium iron phosphate battery data for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention;

[0025] Figure 9 It is a prediction performance diagram for the method for estimating the state of health of power battery in a battery swapping station based on small data samples provided in Embodiment 1 of the present invention. Detailed implementation manners

[0026] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; all other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Example 1

[0028] Figure 1 A flow chart of a method for estimating the health status of power batteries in a battery swap station based on small data samples is provided in Example 1 of the present invention. The embodiment of the present invention can be applied to estimating the health status of power batteries in a battery swap station based on small data samples. The method for estimating the health status of power batteries in a battery swap station based on small data samples can be executed by a system for estimating the health status of power batteries in a battery swap station based on small data samples. The system for estimating the health status of power batteries in a battery swap station based on small data samples can be implemented by software and / or hardware, and the embodiment of the present invention does not limit this.

[0029] like Figure 1 and Figure 4 As shown, the method for estimating the health status of a power battery in a battery swap station based on a small data sample provided by an embodiment of the present invention specifically includes the following steps:

[0030] Step 1: Obtain the usage data of the power battery in the battery swap station, use differential thermovoltammetry to analyze the temperature characteristic value, capacity increment and charging time characteristic value, and analyze to obtain the DTV and IC curves;

[0031] It should be noted that the usage data includes but is not limited to battery service time, battery charging current, battery charging voltage, battery charging capacity and battery charging temperature;

[0032] In some embodiments, lithium batteries exhibit various performance characteristics under complex and changeable working conditions, so a single feature often cannot fully reflect its true state and health. Therefore, the present invention uses a multi-feature analysis method to mine the internal information of the battery from multiple angles, including but not limited to temperature, voltage, and current characteristics, to improve the generalization ability of the model and enable the model to better adapt to different usage scenarios;

[0033] Differential thermal voltammetry (DTV) is used to mine temperature features. The calculation formula is as follows:

[0034] Where T is the battery temperature at time t, V is the voltage, and t is the sampling time;

[0035] As an important characteristic analysis method, the capacity increment IC of the power battery is calculated as another characteristic analysis method, through the formula

[0036] Get the capacity increment IC, where Q represents the discharge capacity, I represents the current, V represents the voltage, and t represents the time;

[0037] During the process of calculating the DTV and IC curves, the measured values of temperature, current, and voltage are accompanied by noise, which will cause unwanted variations in the DTV and IC curves, thereby affecting the extraction of features. Therefore, the present invention uses the Savitzky-Golay filtering method to filter the IC curve;

[0038] Obtain the constant current charging time from 3.8V to 3.9V. Through the formula

[0039] Calculate the charging time eigenvalue. Among them, CTV1 represents the charging time when the charging voltage of the power battery reaches 3.8V, CTV2 represents the charging time when the charging voltage of the power battery reaches 3.9V, and CTV is defined as the charging time in the current voltage range;

[0040] It should be noted that considering the long duration of constant current charging and the time cost of data acquisition, the 3.8V - 3.9V charging time window corresponds to the middle stage of charging. At this time, the voltage change is relatively stable, which can effectively reflect the time difference between cycles. Therefore, the present invention selects CTV 3.9V -CTV 3.8V as an effective feature of battery degradation;

[0041] Step 2: Process the features extracted from the original time series data by CNN using BiLSTM to improve the accuracy of lithium-ion battery capacity trajectory prediction. Apply an attention mechanism after the LSTM layer to dynamically select relevant hidden states in the entire time series. Choose to use transfer learning to solve the problem of limited data in the target domain. In the pre-training stage with sufficient data, use Bayesian optimization to optimize the model hyperparameters and develop a neural network model;

[0042] Use CNN to extract features from the original time series data. BiLSTM consists of two layers of LSTM. Among them, the predicted system state references the output sequence from the bidirectional incoming LSTM layer, and the prediction results are merged and assigned to the next LSTM layer;

[0043] After the second LSTM layer, the final prediction is jointly determined by forward and backward propagation. Choose to use BiLSTM to process the features extracted from the original time series data by CNN to improve the accuracy of lithium-ion battery capacity trajectory prediction;

[0044] By applying an attention mechanism after the LSTM layer to dynamically select relevant hidden states in the entire time series and focus on the relevant states across the time series;

[0045] For the output H = [h1, h2,…,hT] of the hidden layer of LSTM, through the formula:

[0046]

[0047] The attention weights are calculated where T represents the number of outputs of the hidden layer, represents the state vector at time step t, and respectively represent and the exponential functions of 、 and , representing the weight matrices to be learned;

[0048] After obtaining the weights, the weighted sum is calculated through the formula to obtain c, and then the final result is obtained through the fully connected layer by the formula where c represents the relevant hidden state, represents the weight matrix, and b represents the bias vector;

[0049] The problem of limited data in the target domain is solved by using transfer learning, including two main steps: the pre-training stage using source domain data and the fine-tuning stage using target domain data;

[0050] The goal of the fine-tuning stage is to adjust the parameters of the model to better adapt to the specific degradation mode of the target domain; the transfer learning fine-tuning strategy selected in the present invention is as Figure 5 shown, and only the parameters of BiLSTM are opened for training in the fine-tuning stage;

[0051] It should be noted that BiLSTM, as the core of the time series prediction in this framework, needs to be retrained under different battery operating conditions and different battery chemical compositions, and can adapt to the battery degradation mode of the target domain in the case of small samples. At the same time, only opening the parameters of the BiLSTM block for training reduces the complexity of model training in the fine-tuning stage;

[0052] In the pre-training stage with sufficient data, Bayesian optimization is used to optimize the hyperparameters of the model;

[0053] By using Gaussian processes to model the target, the parameter ranges of Bayesian optimization are shown in Table (1) below:

[0054] Table (1) shows the parameter ranges of Bayesian optimization corresponding to different hyperparameters;

[0055]

[0056] The battery public dataset NASA is used for pre-training, and then only the first 10% of the data of the target battery to be predicted is used to fine-tune the network to achieve the prediction of the SOH of the target battery;

[0057] Exemplarily, the Pearson correlation coefficient, as a linear correlation index, is widely used. Its definition for variables X and Y is as follows:

[0058] Among them, is denoted as the Pearson correlation coefficient, and respectively represent the average values of the eigenvalue and the battery capacity value; n represents the sample size; the range of the correlation coefficient is from -1 to 1, where values close to 1 or -1 indicate a strong relationship, and values close to 0 indicate a weak relationship; Figure 6 shows the correlation coefficients between the features extracted by the multi - feature extraction method and the battery capacity; finally, 7 features with a correlation coefficient above 0.8 are selected as the feature inputs for predicting the battery capacity; for the small - sample prediction performance, in the verification stage of the present invention, the model was first pre - trained, and then fine - tuned using the first 10% of the data of the target monomer in the fine - tuning stage. The prediction results are as shown in Figure 7 which are the prediction results of the SOH of ternary lithium batteries under different temperature conditions. In Figure 7 , Figure (a) corresponds to the - 20°C temperature condition, Figure (b) corresponds to the - 10°C temperature condition, Figure (c) corresponds to the 0°C temperature condition, and Figure (d) corresponds to the 25°C temperature condition; it can be seen that under different small - sample data predictions, the present invention has achieved excellent prediction results; Figure 8 are the results obtained on the lithium iron phosphate battery dataset; in the case where both the chemical composition of the battery electrode and the operating conditions are different, the model still has an RMSE error of 0.025 Ah, further proving the generalization of the proposed method; for the prediction performance of different features, in order to comprehensively verify the effectiveness of the proposed multi - feature fusion method and different feature extraction strategies, a series of comparative experiments were designed to analyze the performance of these methods on the experimental dataset, as shown in Figure 9 ; Table (2) further summarizes the key numerical values of the experimental results; the experimental results show that although the single - feature extraction method can capture some aging patterns, its performance has certain limitations when dealing with complex operating conditions and diverse data features; while the proposed multi - feature fusion method significantly improves the overall prediction ability of the model and shows the best prediction effect under all operating conditions; this indicates that the multi - feature fusion method can effectively combine the advantages of different features, more comprehensively characterize the battery aging characteristics, and provide a more reliable solution for complex prediction tasks.

[0059] Table (2) shows the numerical values of the experimental results corresponding to different operating - condition temperatures;

[0060]

[0061] The technical solution of the present invention is as follows: In the embodiments of the present invention, by obtaining the usage data of the power batteries in the battery swapping station, the temperature characteristic values, capacity increment and charging time characteristic values are obtained through differential thermovoltammetry analysis, and the DTV and IC curves are obtained through analysis. By using BiLSTM to process the features extracted from the original time series data by CNN, the prediction accuracy of the lithium-ion battery capacity trajectory is improved. After the LSTM layer, an attention mechanism is applied to dynamically select relevant hidden states in the entire time series, and transfer learning is selected to solve the problem of limited data in the target domain. In the pre-training stage with sufficient data, Bayesian optimization is used to optimize the model hyperparameters; realizing the prediction of the battery health state under small sample data, ensuring the generalization of the method, being applicable to the management of in-vehicle power batteries in the battery swapping station, and achieving accurate prediction in the case of incomplete battery usage data in the battery swapping station, ensuring the safe and reliable operation of the battery swapping station, and contributing to the further promotion of vehicle electrification and the realization of carbon neutrality. The prediction accuracy of the lithium-ion battery capacity trajectory is improved through multi-feature analysis and complex machine learning models, and the flexibility and adaptability of the model are enhanced.

[0062] Embodiment 2

[0063] As Figure 2 shown, based on the charging and maintenance data of the power battery in the battery swapping station obtained in Embodiment 1, the method for estimating the health state of the power battery in the battery swapping station based on small data samples provided by the embodiments of the present invention specifically includes the following steps:

[0064] Step 3: Obtain the power battery data, analyze and process it to generate a discharge duration and number curve, analyze to obtain the discharge characteristic values, calculate to obtain the temperature characteristic values, capacity increment and charging time characteristic values of the power battery, analyze the proportion of abnormal data of the power battery, analyze the charge and discharge conditions of the power battery to obtain the charge and discharge state values, comprehensively analyze the data to obtain the health characteristic values, and perform classification analysis;

[0065] It should be noted that the preset voltage is the power battery voltage set by those skilled in the art of the present invention to obtain more characteristic data for analyzing the charge and discharge state of the power battery. The power battery data includes but is not limited to the duration of continuous discharge reaching the preset voltage, the number of discharges of the power battery, the temperature, voltage and current of the power battery, etc.;

[0066] In some embodiments, based on the obtained aging voltage change curve of the power battery, the duration of continuous discharge of the power battery reaching the preset voltage and the number of discharges of the power battery are obtained. Taking the number of discharges of the power battery as the abscissa and the duration of continuous discharge of the power battery reaching the preset voltage as the ordinate, an X-Y two-dimensional coordinate system is established to generate a discharge duration and number curve;

[0067] Set the discharge duration and number curve as and through the formula

[0068] The calculated curvature , where the obtained curvatures are all non - negative. Obtain the curvatures of all points corresponding to the number of discharge times, and through the formula

[0069] calculate the total curvature corresponding to all discharge times, where n is the total number of discharges of the kinetic energy battery;

[0070] Perform a ratio process on the obtained total curvature and the corresponding total number of discharges to obtain the curve average curvature QL, and analyze the change range of the discharge duration between the discharge times of the power battery;

[0071] Obtain the total number of discharges of the power battery when the power battery continuously discharges to reach the preset voltage and simultaneously reaches the preset shortest duration. Perform a ratio process on the preset shortest duration and the total number of discharges of the power battery to obtain the discharge loss value SZ;

[0072] It should be noted that the preset shortest duration is the standard discharge duration set by the professionals in the technical field of the present invention according to historical experience for judging the aging degree of the power battery when the power battery continuously discharges to reach the preset voltage;

[0073] Comprehensively analyze the obtained curve average curvature QL and discharge loss value SZ, and through the formula

[0074] calculate the discharge characteristic value FT, where a, b, and c are all preset correlation coefficients, a takes the value of 0.137, b takes the value of 0.22, and c takes the value of 1.724;

[0075] It should be noted that the discharge characteristic value FT is calculated from the curve average curvature QL and the discharge loss value SZ. The smaller the curve average curvature QL, the more stable the change range of the discharge duration between the discharge times of the power battery, and the more stable the aging process of the power battery. The smaller the discharge loss value SZ, the more the total number of discharges of the power battery when the power battery continuously discharges to reach the preset voltage and simultaneously reaches the preset shortest duration, and the more times the power battery is used before aging. Comprehensively analyze the characteristic performance of the battery discharge process;

[0076] Calculate and obtain the temperature characteristic value and capacity increment of the power battery. Compare the obtained temperature characteristic value with the preset temperature characteristic threshold to obtain the number of temperature characteristic values greater than the temperature characteristic threshold;

[0077] Perform a ratio process on the number of temperature characteristic values greater than the temperature characteristic threshold and the total number of obtained temperature characteristic values to obtain the temperature anomaly ratio;

[0078] Compare the obtained capacity increment with a preset capacity increment threshold to obtain the number of capacity increments less than the capacity increment threshold;

[0079] Ratio-process the number of capacity increments less than the capacity increment threshold to the total number of obtained capacity increments to obtain the proportion of abnormal capacity;

[0080] Perform weighted summation on the obtained proportion of temperature anomalies and the proportion of abnormal capacity to obtain the proportion of abnormal data YC;

[0081] Obtain the charging time characteristic value CTV of the power battery, subtract the obtained charging time characteristic value CTV from the charging time standard value and take the absolute value to obtain the charging time deviation value, and ratio-process the charging time deviation value to the charging time standard value to obtain the charging time deviation ratio PC;

[0082] Comprehensively analyze the discharge characteristic value FT and the charging time deviation ratio PC, through the formula

[0083] Calculate to obtain the charge and discharge state value CF, where d and e are preset correlation coefficients, the value of d is 1.127, and the value of e is 1.879;

[0084] It should be noted that the charge and discharge state value CF is calculated through the discharge characteristic value FT and the charging time deviation ratio PC. The larger the discharge characteristic value FT, the better the discharge state of the power battery. The smaller the charging time deviation ratio PC, the smaller the deviation of the charging time characteristic value CTV relative to the charging time standard value, indicating that the charging state of the power battery is better, which is conducive to analyzing the basis for the health state of the power battery;

[0085] Comprehensively analyze the obtained proportion of abnormal data YC and the charge and discharge state value CF, through the formula

[0086] Calculate to obtain the health characteristic value JK, where and are preset proportionality coefficients, the value of is 0.417,

[0087] Compare the obtained health characteristic value JK with a preset health characteristic threshold JK0, and analyze and divide the reasonable and useful power batteries as data basis;

[0088] Specifically, if the health characteristic value JK is greater than or equal to the preset health characteristic threshold JK0, it indicates that the power battery data is suitable for the model training data of normal aging of the power battery, and mark the corresponding power battery data as normal aging training data;

[0089] If the health feature value JK is less than the preset health feature threshold JK0, it indicates that the power battery data is suitable for the training data of the abnormal aging model of the power battery, and the corresponding power battery data is marked as abnormal aging data;

[0090] The technical solution of the embodiment of the present invention is as follows: Obtain the duration of the power battery continuously discharging to the preset voltage and the number of discharges of the power battery, analyze and process to generate a curve, analyze to obtain the discharge feature value, synchronously obtain the temperature, voltage and current of the power battery, calculate to obtain the temperature feature value, capacity increment and charging time feature value of the power battery, analyze the proportion of abnormal data of the power battery, analyze the charge and discharge conditions of the power battery to obtain the charge and discharge state value, comprehensively analyze the data to obtain the health feature value, and perform classification analysis; At the same time, monitor the temperature, voltage and current of the power battery to ensure the integrity and accuracy of the data, integrate the discharge feature value FT and the charging time deviation ratio PC to obtain the overall charge and discharge performance score, and based on the abnormal data proportion YC and the charge and discharge state value CF, obtain a comprehensive index for measuring the battery health. According to whether the health feature value JK exceeds the preset threshold JK0, automatically distinguish two different aging modes, which is convenient for distinction and understanding, and supports the construction of a more accurate power battery SOH prediction model by machine learning algorithms.

[0091] Embodiment 3

[0092] Refer to Figure 3 Moreover, the embodiment of the present invention also provides a computer device 3, including: a memory 302, a processor 301, and a computer program 303 stored on the memory 302. When the computer program 303 is executed on the processor 301, it implements the method for estimating the health state of the power battery of the swapping station based on small data samples as described in any one of the above methods.

[0093] The computer device 3 may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server; the computer device 3 may include, but is not limited to, a processor 301 and a memory 302, which can be understood by those skilled in the art;

[0094] Figure 3 These are only examples of the computer device 3 and do not constitute a limitation to the computer device 3. It may include more or fewer components than shown in the figure, or combine some components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0095] The so-called processor 301 may be a central processing unit (CPU), and this processor 301 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.; the general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0096] In some embodiments, the memory 302 may be an internal storage unit of the computer device 3, such as the hard disk or memory of the computer device 3. In some other embodiments, the memory 302 may also be an external storage device of the computer device 3, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the computer device 3; further, the memory 302 may also include both the internal storage unit and the external storage device of the computer device 3. The memory 302 is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program, etc.; the memory 302 may also be used to temporarily store data that has been output or is to be output.

[0097] Embodiment 4

[0098] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, it implements the method for estimating the health state of the power battery of a battery swapping station based on a small data sample as described in any one of the above methods.

[0099] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above embodiment methods of this application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the photographing device / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0100] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0101] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0102] In the embodiments disclosed in this application, it should be understood that the disclosed device / terminal device and method can be implemented in other ways. For example, the device / terminal device embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another

[0103] point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be in an electrical, mechanical or other form.

[0104] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0106] The above has described in detail an embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. A method for estimating the state of health of power batteries in a battery swapping station based on small data samples, characterized in that, Including the following steps: By obtaining the usage data of the power batteries in the battery swapping station, analyzing to obtain the temperature characteristic value, capacity increment and charging time characteristic value, and analyzing to obtain the DTV and IC curves; By using BiLSTM to process the features extracted from the original time series data by CNN, and applying the attention mechanism after the LSTM layer to dynamically select relevant hidden states in the whole time series; Select to use transfer learning to solve the problem of limited data in the target domain. In the pre-training stage with sufficient data, use Bayesian optimization to optimize the model hyperparameters and develop a neural network model; By analyzing the DTV and IC curves to obtain the discharge characteristic value, calculating to obtain the temperature characteristic value, capacity increment and charging time characteristic value of the power battery, and analyzing to obtain the proportion of abnormal data of the power battery; Analyze the charge and discharge conditions of the power battery to obtain the charge and discharge state value, comprehensively analyze the data to obtain the health characteristic value, and conduct classification analysis; The specific process of obtaining the health characteristic value is as follows: Comprehensively analyze the obtained proportion of abnormal data YC and the charge and discharge state value CF, through the formula ； Calculate the obtained health eigenvalue JK, where, and are preset proportionality coefficients; Compare the obtained health characteristic value JK with the preset health characteristic threshold JK0, and analyze and divide the reasonable and useful power batteries as data basis; Specifically, if the health characteristic value JK is greater than or equal to the preset health characteristic threshold JK0, it indicates that the power battery data is suitable for the model training data of normal aging of the power battery, and mark the corresponding power battery data as normal aging training data; If the health characteristic value JK is less than the preset health characteristic threshold JK0, it indicates that the power battery data is suitable for the model training data of abnormal aging of the power battery, and mark the corresponding power battery data as abnormal aging data; The specific process of obtaining the proportion of abnormal data is as follows: Calculate to obtain the temperature characteristic value and capacity increment of the power battery, compare the obtained temperature characteristic value with the preset temperature characteristic threshold, and obtain the number of temperature characteristic values greater than the temperature characteristic threshold; Perform a ratio process on the number of temperature characteristic values greater than the temperature characteristic threshold and the total number of obtained temperature characteristic values to obtain the temperature anomaly ratio; Compare the obtained capacity increment with the preset capacity increment threshold, and obtain the number of capacity increments less than the capacity increment threshold; Perform a ratio process on the number of capacity increments less than the capacity increment threshold and the total number of obtained capacity increments to obtain the capacity anomaly ratio; Perform a weighted sum on the obtained temperature anomaly ratio and capacity anomaly ratio to obtain the proportion of abnormal data YC; The specific process of obtaining the charge and discharge state value is as follows: Obtain the charging time characteristic value CTV of the power battery, perform a difference process on the obtained charging time characteristic value CTV and the charging time standard value and take the absolute value to obtain the charging time deviation value, and perform a ratio process on the charging time deviation value and the charging time standard value to obtain the charging time deviation ratio PC; Comprehensively analyze the discharge characteristic value FT and the charging time deviation ratio PC, through the formula ； Calculate to obtain the charge and discharge state value CF, where d and e are preset correlation coefficients.

2. The method for estimating the health state of power batteries in a battery swapping station based on small data samples according to claim 1, wherein The specific process of obtaining the temperature characteristic value, capacity increment and charging time characteristic value is as follows: Differential Thermal Voltammetry (DTV) is used to extract temperature features, and the calculation formula is as follows: ； Where, T is the battery temperature at time t, V is the voltage, and t is the sampling time; Calculating the capacity increment IC of the power battery as another feature analysis method, through the formula: ； Where Q represents the discharge capacity, I represents the current, V represents the voltage, and t represents the time; During the calculation of the DTV and IC curves, the measured values of temperature, current, and voltage are accompanied by noise, which will cause unwanted changes in the DTV and IC curves, thus affecting the feature extraction; the Savitzky-Golay filtering method is used to filter the IC curve; Obtain the constant current charging time from 3.8V to 3.9V, through the formula ； Calculate the charging time eigenvalue, where CTV1 represents the charging time when the charging voltage of the power battery reaches 3.8V, CTV2 represents the charging time when the charging voltage of the power battery reaches 3.9V, and CTV is defined as the charging time in the current voltage range.

3. The method for estimating the health state of power batteries in a battery swapping station based on small data samples according to claim 1, wherein The specific process of processing features is as follows: Use CNN to extract features from the original time series data. BiLSTM consists of two layers of LSTM. Among them, the predicted system state references the output sequence from the bidirectional incoming LSTM layer, and the prediction results are merged and assigned to the next LSTM layer; After the second LSTM layer, the final prediction is jointly determined by forward and backward propagation. Choose to use BiLSTM to process the features extracted by CNN from the original time series data to improve the accuracy of lithium-ion battery capacity trajectory prediction; By applying an attention mechanism after the LSTM layer to dynamically select relevant hidden states throughout the time series, focusing on relevant states across the time series; For the output H = [h1, h2, …, hT] of the hidden layer of LSTM, through the formula: ； Calculated attention weights , where T represents the number of outputs of the hidden layer, represents the state vector at time step t, and respectively represent and exponential functions of, , and represent weight matrices to be learned; After obtaining the weights, the weighted sum is calculated through the formula to get c, and then the fully connected layer uses the formula to obtain the final result, where c represents the relevant hidden state, W represents the weight matrix, and b represents the bias vector.

4. The method for estimating the state of health of power batteries in a battery swapping station based on small data samples according to claim 1, wherein, The specific steps of the transfer learning are as follows: Solve the problem of limited data in the target domain by using transfer learning, including two main steps: the pre-training stage using source domain data and the fine-tuning stage using target domain data; The goal of the fine-tuning stage is to adjust the parameters of the model to better adapt to the specific degradation mode of the target domain; select the transfer learning fine-tuning strategy, and only open the parameters of BiLSTM for training during the fine-tuning stage; In the pre-training stage with sufficient data, use Bayesian optimization to optimize the model hyperparameters by using Gaussian processes to model the target; Use the battery public dataset NASA for pre-training, and then only use the first 10% of the data of the target battery to fine-tune the network to achieve the prediction of the SOH of the target battery.

5. The method for estimating the health state of power batteries in a battery swapping station based on small data samples according to claim 1, wherein The specific process of obtaining the discharge eigenvalue is as follows: Comprehensively analyze the obtained curve average curvature QL and discharge loss value SZ, through the formula ； Calculate the discharge eigenvalue FT, where a, b, and c are all preset correlation coefficients.

6. The method for estimating the state of health of power batteries in a battery swapping station based on small data samples according to claim 5, wherein The specific process of obtaining the curve average curvature is as follows: Based on the obtained aging voltage change curve of the power battery, obtain the duration for which the power battery continuously discharges until it reaches the preset voltage and the number of times the power battery discharges. Taking the number of times the power battery discharges as the abscissa and the duration for which the power battery continuously discharges until it reaches the preset voltage as the ordinate, establish an X-Y two-dimensional coordinate system and generate a discharge duration-times curve; Set the discharge duration times curve as , through the formula ； Calculated curvature , where the obtained curvatures are all non-negative, and the curvatures corresponding to all discharge times are obtained through the formula ； Calculate the total curvature corresponding to all discharge times, where n is the total number of discharges of the kinetic energy battery; Perform a ratio process on the obtained total curvature and the corresponding total number of discharges to obtain the curve average curvature QL, and analyze the variation range of the discharge duration between the discharge times of the power battery.

7. The method for estimating the state of health of power batteries in a battery swapping station based on small data samples according to claim 5, characterized in that, The specific process for obtaining the discharge loss value is as follows: Obtain the total number of discharges of the power battery that reaches the preset shortest duration while continuously discharging until it reaches the preset voltage, and perform a ratio process on the preset shortest duration and the total number of discharges of the power battery to obtain the discharge loss value SZ.

Images