Battery health state prediction method and device and storage medium

Abstract

The invention relates to a battery health state prediction method and device and a storage medium, relates to the technical field of battery detection, and aims to at least solve the problem that a battery health degree prediction result is inaccurate due to single data dimension and lack of training samples. The method comprises the following steps: determining actual measurement data generated in the operation process of a target battery before the current time; according to third label data generated by the actual measurement data, optimizing model parameters of the trained preset model to obtain an optimized target model; the preset model is obtained by training in a supervised learning training mode based on first label data generated by performing data labeling according to test data of the target battery and second label data generated by performing data labeling according to historical experiment data; and inputting the current operation parameters of the target battery at the current time into the target model to obtain the current health degree.

Description

Technical Field

[0001] The present disclosure relates to the technical field of battery detection, and particularly to a method, device, and storage medium for predicting the state of health of a battery. Background Art

[0002] With the rapid development of fields such as electric vehicles and energy storage systems, the battery, as the core energy storage device, directly affects the reliability and safety of the entire system in terms of its performance and state of health. The State of Health (SOH) of a battery is an important indicator for measuring the degree of battery performance degradation. Accurately evaluating the battery SOH is of great significance for predicting battery life, optimizing battery usage strategies, and avoiding battery failures.

[0003] During the training process of unsupervised learning in the current battery prediction model, it relies on a single data source (such as voltage and current), fails to fully utilize the complementarity of multi-source data, and the model's prediction of the health degree is inaccurate. When using supervised learning for training, a large amount of training sample data is required, but in some new industries (such as electric vehicles and energy storage systems), sufficient sample data cannot be collected, resulting in the model obtained by the supervised training method also having inaccurate prediction results for the health degree. Summary of the Invention

[0004] The present invention provides a method, device, and storage medium for predicting the state of health of a battery to at least solve the problem of inaccurate prediction results of the battery health degree due to single data dimension and lack of training samples. The technical solution of the present invention is as follows:

[0005] According to the first aspect of the embodiments of the present invention, a method for predicting the state of health of a battery is provided. The method includes: determining the actual measurement data generated during the operation of the target battery before the current time; the actual measurement data includes actual operating parameters and actual health degree; the actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual charge and discharge cycle times of the target battery; optimizing the model parameters of a preset model that has been trained and completed according to the third label data generated from the actual measurement data to obtain an optimized target model; the preset model is trained through a supervised learning training method based on first label data and second label data. The first label data is generated by data labeling according to the test data of the target battery; the second label data is generated by data labeling according to historical experimental data; the historical experimental data is data generated by a historical battery of the same type as the target battery and that has been put into use; inputting the current operating parameters of the target battery at the current time into the target model to obtain the current health degree.

[0006] In one implementation, before determining the actual measurement data generated during the operation of the target battery before the current time, the method includes: conducting a battery standardization test on the target battery to obtain test data; the test data includes the test voltage, test current, test temperature, test operation time, test charge-discharge cycle times, and test health status of the target battery; according to the test data, fitting and optimizing the first model parameters of the empirical model constructed by the capacity attenuation rate and the capacity under temperature and charge cycle times to obtain the first target empirical model; the empirical model is constructed based on the Arrhenius formula of temperature change; generating first simulation data according to the first target empirical model; labeling the first health status and the corresponding test operation parameters in the first simulation data and the test data according to the correlation mapping relationship between the capacity attenuation rate and the health status to obtain the first labeled data.

[0007] By obtaining multi-dimensional data of the target battery and utilizing the complementarity between the multi-dimensional data, it is ensured to obtain accurate data support in a small number of tests. Based on this, an empirical model is established with a small number of tests, shortening the experimental time-consuming and improving the timeliness.

[0008] In one implementation, before determining the actual measurement data generated during the operation of the target battery before the current time, the method includes: labeling the second health status and the corresponding historical operation parameters in the historical experimental data to obtain the second labeled data.

[0009] Utilizing the historical experimental data of the same type of battery as the target battery increases the sufficient amount of data for the training process.

[0010] In one implementation, labeling the second health status of the historical battery according to the historical experimental data includes: determining the data interval of the historical experimental data; determining that the data interval in the historical experimental data is less than the preset interval, and using the interpolation method to insert the second simulation data into the historical experimental data; labeling the second health status and the corresponding historical operation parameters in the historical experimental data and the second simulation data to obtain the second labeled data.

[0011] In one implementation, the method further includes: determining that the data interval in the historical experimental data is greater than or equal to the preset interval, and using the extrapolation method to insert the third simulation data into the historical experimental data; labeling the second health status and the corresponding historical operation parameters in the historical experimental data and the third simulation data to obtain the second labeled data.

[0012] Based on the existing battery-related data, using the interpolation method or the extrapolation method to estimate the health status labeled data of the battery at the current or a future moment to ensure the complete cycle of the data, thereby providing more comprehensive and complete data support for model training.

[0013] In one implementation, the third tag data generated according to the actual measurement data includes: fitting and optimizing the second model parameters of the first target empirical model according to the actual measurement data to obtain a second target empirical model; generating fourth simulation data according to the second target empirical model; and tagging the third health status and corresponding operating parameters in the fourth simulation data and the actual measurement data according to the association mapping relationship between the capacity attenuation rate and the health status to obtain the third tag data.

[0014] In one implementation, the method further includes: when the current health status is greater than the preset health status, determining that the target battery is in a healthy state; when the current health status is less than or equal to the preset health status, determining that the target battery is in an unhealthy state.

[0015] In one implementation, the method further includes: obtaining multiple current health statuses in real time according to a preset period; determining the total number of charge and discharge cycles when reaching the preset health status according to the trend of the health status characterized by the multiple current health statuses; and determining the service life of the target battery according to the total number of charge and discharge cycles and the current number of charge and discharge cycles.

[0016] According to the second aspect of the embodiments of the present invention, there is provided a device for predicting the health status of a battery, the device including: a determination unit configured to determine actual measurement data generated during the operation of the target battery before the current time; the actual measurement data includes actual operating parameters and actual health status; the actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual number of charge and discharge cycles of the target battery.

[0017] A generation unit is configured to optimize the model parameters of a preset model that has been trained according to the third tag data generated according to the actual measurement data to obtain an optimized target model; the preset model is trained by a supervised learning training method based on the first tag data and the second tag data, the first tag data is generated by data tagging according to the test data of the target battery; the second tag data is generated by data tagging according to historical experimental data; the historical experimental data is data generated by a historical battery of the same type as the target battery and that has been put into use.

[0018] A prediction unit is configured to input the current operating parameters of the target battery at the current time into the target model to obtain the current health status.

[0019] According to the third aspect of the embodiments of the present invention, there is provided a computer-readable storage medium, on which instructions are stored. When the instructions in the computer-readable storage medium are executed by a processor of a device for predicting the health status of a battery, the device for predicting the health status of a battery can execute the method for predicting the health status of a battery as described in the first aspect and any possible implementation manner thereof.

[0020] The technical solutions provided by the embodiments of the present invention at least bring the following beneficial effects: By collecting the actual operating parameters and actual health of the target battery, multi-dimensional data of the actual operation of the battery can be obtained, making full use of the complementarity of the data to understand the battery state more comprehensively and accurately, so as to obtain more accurate data. The preset model is supervised-trained with the test data of the target battery itself and the experimental data of the same type of battery. The test data can specifically characterize the current working conditions of the target battery, and the experimental data of the same type of battery as the target battery is introduced, increasing the amount of data in the training process, thereby improving the accuracy of the preset model. The model parameters of the trained preset model are further optimized to make the optimized target model more consistent with the current working conditions of the target battery, thereby ensuring that the current health obtained based on the target model is more accurate.

[0021] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present disclosure, and are used together with the specification to explain the principles of the present disclosure, and do not constitute an improper limitation of the present disclosure.

[0023] Figure 1 is a schematic diagram of a battery health state prediction system shown according to an exemplary embodiment;

[0024] Figure 2 is a flowchart of a battery health state prediction method shown according to an exemplary embodiment Figure 1 ;

[0025] Figure 3 is a flowchart of a battery health state prediction method shown according to an exemplary embodiment Figure 2 ;

[0026] Figure 4 is a block diagram of a battery health state prediction device shown according to an exemplary embodiment;

[0027] Figure 5 is a schematic diagram of an electronic device shown according to an exemplary embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] In order to enable those of ordinary skill in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0029] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present disclosure are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order different from those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0030] Before introducing the battery health state prediction method provided by the embodiments of the present application in detail, the application scenarios and implementation frameworks involved in the embodiments of the present application will be briefly introduced.

[0031] First, the application scenarios involved in the present application will be briefly introduced.

[0032] With the rapid development of fields such as electric vehicles and energy storage systems, the battery, as the core energy storage device, its performance and health state directly affect the reliability and safety of the entire system. Battery health is an important indicator to measure the degree of battery performance degradation. Accurately evaluating battery health is of great significance for predicting battery life, optimizing battery usage strategies, and avoiding battery failures.

[0033] It has been found through research that currently, there are model prediction methods, empirical formula simulation methods, and data-driven methods for predicting battery health. Model prediction refers to establishing an electrochemical model of the battery and using software such as COMSOL to predict battery life, but it requires a large amount of battery material data and preparation data, with a high degree of complexity. It has high accuracy in laboratory applications. However, because the acquisition accuracy of complex working conditions itself cannot meet the model requirements, it is not suitable for evaluation under complex working conditions. The empirical formula simulation method refers to using a test bench to test the relationship between battery health and cycles, temperature, or charge and discharge amount to obtain an empirical formula, thereby estimating battery health. However, a very large number of tests need to be carried out for each type of battery, which is time-consuming, and at the same time, it cannot reflect the situation of the battery in complex actual use, especially the attenuation at the end of the battery life. The data-driven method refers to collecting the historical operation data of the battery and using machine learning algorithms to establish a battery health evaluation model. This type of method is generally unsupervised learning and depends on a single data source (such as voltage, current), fails to make full use of the complementarity of multi-source data, has low accuracy in the initial stage of vehicle operation, requires a large amount of data, and at the same time, model training requires a large amount of high-quality data, and the data acquisition cost is relatively high.

[0034] In view of the above problems, the present application proposes a method for predicting the state of health of a battery. By collecting the actual operating parameters and the actual state of health of the target battery, multi-dimensional data of the actual operation of the battery is obtained, making full use of the complementarity of the data to understand the battery state more comprehensively and accurately, so as to obtain more accurate data. The preset model is supervised-trained using the test data of the target battery itself and the experimental data of the same type of battery. The test data can specifically characterize the current working conditions of the target battery, and the experimental data of the same type of battery as the target battery is introduced, increasing the amount of data in the training process, thereby improving the accuracy of the preset model. The model parameters of the preset model that has been trained are further optimized so that the optimized target model fits better with the current working conditions of the target battery, thereby ensuring that the current state of health obtained based on the target model is more accurate.

[0035] Secondly, a brief introduction to the implementation architecture involved in the present application is given below.

[0036] Figure 1 It is a schematic diagram of a battery state of health prediction system provided by the present application. As Figure 1 shown, the battery state of health prediction system includes a data acquisition module 11, a data preprocessing module 12, a feature extraction module 13, an empirical model label generation module 14, a model training module 15, a state of health evaluation module 16, a result output module 17, and a cloud deployment and synchronization module 18.

[0037] The above data acquisition module 11, data preprocessing module 12, feature extraction module 13, empirical model label generation module 14, model training module 15, state of health evaluation module 16, result output module 17, and cloud deployment and synchronization module 18 are communicatively connected via a wired network or a wireless network.

[0038] The data acquisition module 11 is configured to collect the actual measurement data generated during the operation of the target battery before the current time.

[0039] Among them, the actual measurement data includes actual operating parameters and actual state of health; the actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual charge and discharge cycle times of the target battery.

[0040] The data preprocessing module 12 is configured to preprocess the collected data, including cleaning, denoising, and normalization.

[0041] The feature extraction module 13 is configured to extract the feature parameters related to the state of health of the battery from the preprocessed data, including the capacity attenuation rate, internal resistance change rate, etc.

[0042] The empirical model label generation module 14 is configured to optimize the fitting of the first model parameters of the empirical model constructed based on the capacity decay rate, temperature, and capacity under the number of charge cycles according to the test data, and obtain the first target empirical model.

[0043] Among them, the empirical model is constructed based on the Arrhenius formula for temperature change. According to the correlation mapping relationship between the capacity decay rate and the health degree, the first health degree and the corresponding test operation parameters in the first simulation data and the test data are labeled to obtain the first labeled data.

[0044] The model training module 15 is configured to optimize the model parameters of the pre-trained preset model according to the third labeled data generated from the actual measurement data, and obtain the optimized target model.

[0045] Among them, the preset model is trained by a supervised learning training method based on the first labeled data and the second labeled data; the first labeled data is generated by data labeling according to the test data of the target battery; the second labeled data is generated by data labeling according to the historical experimental data; the historical experimental data is the data generated by the historical battery of the same type as the target battery and already in use.

[0046] The health degree evaluation module 16 is configured to input the current operation parameters of the target battery at the current time into the target model to obtain the current health degree evaluation result.

[0047] The result output module 17 is configured to output whether the target battery is in a healthy state or an unhealthy state according to the health degree evaluation result.

[0048] The cloud deployment and synchronization module 18 is configured to transmit the extracted characteristic parameters related to the health degree of the target battery to the cloud big data, realize multi-terminal data synchronization and sharing, and apply it to large-scale battery management scenarios.

[0049] The battery health state prediction method provided by the embodiments of the present application can be applied to the battery health state prediction system in the foregoing Figure 1 shown implementation architecture. For the convenience of understanding, the battery health state prediction method provided by the present application will be specifically introduced below with reference to the accompanying drawings.

[0050] Figure 2 is a flowchart of a battery health state prediction method shown according to an exemplary embodiment Figure 1 as Figure 2 shown, and this battery health state prediction method includes the following steps.

[0051] S21, determine the actual measurement data generated during the operation of the target battery before the current time.

[0052] The actual measurement data includes actual operating parameters and actual health status.

[0053] The actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual charge-discharge cycle times of the target battery.

[0054] Real-time data collection is performed on the target battery to determine the changes of the target battery during actual use, providing accurate data support for the optimization of the target model and the dynamic update of the health status results.

[0055] S22. Optimize the model parameters of the pre-trained preset model according to the third label data generated from the actual measurement data to obtain the optimized target model.

[0056] The preset model is trained through a supervised learning training method based on the first label data and the second label data.

[0057] The first label data is generated by data labeling according to the test data of the target battery.

[0058] The second label data is generated by data labeling according to historical experimental data.

[0059] The historical experimental data is data generated by historical batteries of the same type as the target battery that have been put into use.

[0060] Optionally, the third label data generated according to the actual measurement data includes: fitting and optimizing the second model parameters of the first target empirical model according to the actual measurement data to obtain the second target empirical model, generating the fourth simulation data according to the second target empirical model, and labeling the third health status and the corresponding operating parameters in the fourth simulation data and the actual measurement data according to the correlation mapping relationship between the capacity attenuation rate and the health status to obtain the third label data.

[0061] The first target empirical model is obtained by performing a battery standardization test on the target battery to obtain test data before determining the actual measurement data generated during the operation of the target battery before the current time, and fitting and optimizing the first model parameters of the empirical model constructed by the capacity attenuation rate and the capacity under temperature and charge cycle times according to the test data.

[0062] The second target empirical model represents fitting and optimizing based on the first target empirical model using the current actual measurement data.

[0063] The fourth simulated data represents the battery health-related data obtained by inputting the actual measurement data into the second target empirical model. In one implementation, during the actual use of the battery, the third label data of the battery health is dynamically updated through the actual measurement data, and the preset model is periodically tested or calibrated to obtain the target model, so as to ensure the accuracy of the battery health estimated by the target model.

[0064] Exemplarily, after every 100 cycles, a capacity test is performed to update the battery health label.

[0065] S23, input the current operating parameters of the target battery at the current time into the target model to obtain the current health.

[0066] Optionally, if the current health is greater than the preset health, it is determined that the target battery is in a healthy state; if the current health is less than or equal to the preset health, it is determined that the target battery is in an unhealthy state.

[0067] The current health represents the actual health result of the target battery at present obtained through the operation of the target model.

[0068] The preset health represents a preset health value as a measurement standard.

[0069] When the health of the target battery reaches or exceeds this standard, it is considered that the target battery is in a healthy state, indicating that the current performance and condition of the target battery are good and it can continue to be used.

[0070] When the health of the target battery does not reach this standard, it is considered that the target battery is in an unhealthy state, indicating that the target battery may have problems such as performance degradation and aging, and it cannot continue to be used as a charging and discharging battery, but can be used as an energy storage device.

[0071] Exemplarily, the preset health is set to 80%. If the current health of the target battery is detected to be 85% at present, since 85% is greater than 80%, it can be determined that the target battery is in a healthy state. If the current health of the target battery is detected to be 65% at present, since 65% is less than 80%, it can be determined that the target battery is in an unhealthy state.

[0072] Optionally, according to the preset period, multiple current healths are obtained in real time. According to the health trend characterized by the multiple current healths, the total number of charge and discharge cycles when the preset health is reached is determined, and according to the total number of charge and discharge cycles and the current number of charge and discharge cycles, the service life of the target battery is determined.

[0073] The health trend represents the change trend of the battery health over time or usage conditions by analyzing the data characterized by multiple current healths.

[0074] By collecting multiple pieces of index data that can characterize the battery health, analyzing this data to obtain the changing trend of the battery health, making a prediction based on the changing trend, estimating the total number of charge and discharge cycles when the battery reaches the preset health, and obtaining the number of charge and discharge cycles that the target battery can still be used by the difference between the determined total number of charge and discharge cycles and the number of charge and discharge cycles that the battery has currently completed, so as to measure the service life of the battery.

[0075] Exemplarily, by analyzing the battery health characterizations such as the capacity and internal resistance of the battery, it is predicted that the total number of charge and discharge cycles when the battery reaches the preset health is 1000 times, and the current battery has completed 300 charge and discharge cycles, then the service life of this battery is 700 charge and discharge cycles.

[0076] In one implementation manner, as Figure 3 shown, the process of determining the preset model in the above step S22 is specifically implemented through the following steps S221 to S224.

[0077] S221, determine the test data.

[0078] Before determining the actual measurement data generated during the operation of the target battery before the current time, perform a battery standardization test on the target battery to obtain the test data, and the test data includes the test voltage, test current, test temperature, test operation time, test charge and discharge cycle number, and test health of the target battery.

[0079] Specifically, further calculate the change of the battery internal resistance, the cumulative capacity of the battery, etc. based on the test data of the target battery to obtain the multi-dimensional data of the battery, and make full use of the complementarity of the data. By comprehensively analyzing the test data of the target battery, the state and performance of the battery can be understood more comprehensively and accurately. For example, by combining the voltage, current, and temperature data, it can be judged whether the battery is in a normal charge and discharge state and whether there are abnormal situations such as overheating; by combining the internal resistance and the number of charge and discharge cycles, the cumulative capacity and capacity attenuation rate of the battery can be further evaluated, so as to provide a more reliable basis for predicting the battery health.

[0080] S222, generate the first label data for data labeling according to the test data of the target battery.

[0081] The test data is obtained by performing a battery standardization test on the target battery through the above step S221.

[0082] According to the test data, perform fitting optimization on the first model parameters of the empirical model constructed by the capacity attenuation rate and the capacity under temperature and charge cycle number to obtain the first target empirical model, and the empirical model is constructed based on the Arrhenius formula of temperature change.

[0083] In one embodiment, using the data obtained from testing, the relationships between different parameters are found, such as the correlation between the capacity attenuation rate and the number of battery charge and discharge cycles. This relationship is fitted into an empirical formula by mathematical methods. Also, in different cycle tests, the temperature is different and affects the battery capacity attenuation rate. An empirical model is constructed based on the Arrhenius formula for temperature change as shown in the following formula (1).

[0084]

[0085] Where Q loss is the percentage of capacity loss; B is the pre-exponential factor; E a is the activation energy; R is the universal gas constant; T is the absolute temperature; z is the power function factor; A h is the ampere-hour throughput, i.e., the cumulative capacity of the number of cycles.

[0086] B is the pre-exponential factor, E a is the activation energy, T is the absolute temperature, and z is the power function factor, which are all different in each re-test. Based on this, the first target empirical model can be iteratively updated using their variation results to make the first target empirical model more accurate.

[0087] According to the first target empirical model, first simulated data is generated. Based on the correlation mapping relationship between the capacity attenuation rate and the health state, the first health state and the corresponding test operation parameters in the first simulated data and the test data are labeled to obtain the first labeled data.

[0088] In one embodiment, the first simulated data generated from the first target empirical model includes the capacity attenuation rate. The correlation mapping relationship between the capacity attenuation rate and the battery health state is determined, as shown in the following formula (2) specifically.

[0089] SOH = 1 - Q loss (2).

[0090] Where SOH is the battery health state, and Q loss is the percentage of capacity loss, also known as the capacity attenuation rate. That is, the lower the capacity attenuation rate, the higher the battery health state.

[0091] Based on the above steps, it can be obtained that by utilizing the complementarity between multi-dimensional data, accurate data support can be ensured in a small number of tests. Further, an empirical model is established using a small number of test data containing multi-dimensional influencing factors, and the health state of the test data is labeled, greatly shortening the experimental time-consuming.

[0092] S223. Generate second labeled data according to the historical experimental data for data labeling.

[0093] The historical experimental data is the data generated by historical batteries of the same type as the target battery and that have been put into use.

[0094] Optionally, label the second health degree and the corresponding historical operation parameters in the historical experimental data to obtain second label data.

[0095] In one implementation, based on the different data interval degrees in the historical experimental data, two different methods are used to estimate the historical data, so as to obtain more accurate second label data.

[0096] First, when it is determined that the data interval degree in the historical experimental data is less than the preset interval degree, interpolation method is used to insert second simulated data into the historical experimental data, and label the second health degree and the corresponding historical operation parameters in the historical experimental data and the second simulated data to obtain second label data.

[0097] The preset interval degree represents a standard value set according to the difference degree between each data point in the historical experimental data.

[0098] The second simulated data is obtained by predicting the data based on the historical experimental data according to the interpolation method.

[0099] It can be understood that the interpolation method utilizes the linear relationship between data. When the interval between data points is small, it means that the change of data is relatively gentle, and the relationship between adjacent data points is approximately linear. At this time, linear interpolation is used. Assuming that the change between two known data points is in a straight line form, through the coordinates of these two points, the value at any position between the two points is estimated according to the linear equation, which can more accurately reflect the actual situation of the data. Therefore, when it is determined that the data interval degree in the historical experimental data is less than the preset interval degree, using the interpolation method to insert second simulated data into the historical experimental data can obtain more accurate second label data.

[0100] Second, when it is determined that the data interval degree in the historical experimental data is greater than or equal to the preset interval degree, extrapolation method is used to insert third simulated data into the historical experimental data, and label the second health degree and the corresponding historical operation parameters in the historical experimental data and the third simulated data to obtain second label data.

[0101] The third simulated data is obtained by predicting the data based on the historical experimental data according to the extrapolation method.

[0102] It is understandable that extrapolation predicts values beyond the known data range based on the trend of known data points. When the interval between data points is large, the change of data may be relatively complex, no longer a simple linear relationship, and there may be some non-linear rules. At this time, using non-linear extrapolation methods, such as polynomial extrapolation, exponential extrapolation, etc., can better capture the change trend of data, so as to make a more reasonable prediction of the values in the unknown region. Therefore, when it is determined that the data interval in the historical experimental data is greater than or equal to the preset interval, and the extrapolation method is used to insert the third simulated data into the historical experimental data to supplement the comprehensiveness of this data, so as to make the second label data result more accurate.

[0103] Exemplarily, the historical experimental data shows that the capacity of the same type of battery decays to 90% after 100 cycles. The second label generation result is that assuming the current cycle number of the battery is 50 times, its health is estimated to be 95%.

[0104] According to the above steps, using the historical experimental data of the same type of battery as the target battery increases the amount of data for the training process. And based on the existing battery historical data, the health label data of the battery at the current or a certain future moment is estimated, and the interpolation method or extrapolation method is used to ensure the complete cycle of the data, so as to provide more intensive and comprehensive data support for model training.

[0105] S224, based on the first label data and the second label data, a preset model is trained through a supervised learning training method.

[0106] In one embodiment, the first label data is generated by data labeling according to the test data in step S222. The second label data is generated by data labeling according to the historical experimental data in step S223.

[0107] Among them, the supervised learning training algorithms include algorithms such as random forest and neural network.

[0108] Based on the first label data and the second label data provides an accurate training target for the supervised learning training of the preset model. At the same time, it provides a variety of data for the preset model, thus significantly improving the evaluation accuracy of the model.

[0109] It can be seen from the above steps that the battery health state prediction method, based on a small amount of accurate test data of the target battery in advance, supplements the experimental data of the historical batteries of the same type as the target battery that have been put into use, ensures the collection and analysis of a large amount of data in a short time, provides sufficient data support for establishing an accurate preset model, and improves the model dynamically by the actual measurement data and real-time working conditions to adjust the preset model to adapt to the changes of the battery in actual use to obtain the target model. Input the current operating parameters of the target battery at the current time into the target model to obtain an accurate health result.

[0110] To implement the above functions, the battery health state prediction device includes the corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should easily realize that, in combination with the algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the manner of hardware or computer software driving hardware 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 the present application.

[0111] The embodiments of the present disclosure also provide a battery health state prediction device as Figure 4 shown, and the device includes: a determination unit 201, a generation unit 202, and a verification unit prediction unit 203.

[0112] The determination unit 201 is configured to determine the actual measurement data generated during the operation of the target battery before the current time; the actual measurement data includes actual operation parameters and actual health; the actual operation parameters include the actual voltage, actual current, actual temperature, actual operation time, and actual charge and discharge cycle times of the target battery;

[0113] The generation unit 202 is configured to optimize the model parameters of the pre-trained preset model according to the third tag data generated from the actual measurement data to obtain an optimized target model; the preset model is obtained by training through a supervised learning training method based on the first tag data and the second tag data. The first tag data is generated by data tagging according to the test data of the target battery; the second tag data is generated by data tagging according to the historical experimental data; the historical experimental data is the data generated by the historical battery of the same type as the target battery and already in use;

[0114] The prediction unit 203 is configured to input the current operation parameters of the target battery at the current time into the target model to obtain the current health.

[0115] In one embodiment, the generating unit 202 is specifically configured to: before determining the actual measurement data generated during the operation of the target battery before the current time, perform a battery standardization test on the target battery to obtain test data, where the test data includes the test voltage, test current, test temperature, test operation time, test charge and discharge cycle times, and test health of the target battery. According to the test data, perform fitting optimization on the first model parameters of the empirical model constructed by the capacity decay rate and the capacity under temperature and charge cycle times to obtain the first target empirical model; the empirical model is constructed based on the Arrhenius formula of temperature change. Generate first simulated data according to the first target empirical model. According to the association mapping relationship between the capacity decay rate and the health, label the first health and the corresponding test operation parameters in the first simulated data and the test data to obtain the first labeled data.

[0116] In one embodiment, the generating unit 202 is specifically configured to: before determining the actual measurement data generated during the operation of the target battery before the current time, label the second health and the corresponding historical operation parameters in the historical experimental data to obtain the second labeled data.

[0117] In one embodiment, the generating unit 202 is specifically configured to: label the second health of the historical battery according to the historical experimental data, including: determining the data interval of the historical experimental data, determining that the data interval in the historical experimental data is less than the preset interval, and using the interpolation method to insert second simulated data into the historical experimental data, and label the second health and the corresponding historical operation parameters in the historical experimental data and the second simulated data to obtain the second labeled data.

[0118] In one embodiment, the generating unit 202 is specifically configured to: determine that the data interval in the historical experimental data is greater than or equal to the preset interval, and use the extrapolation method to insert third simulated data into the historical experimental data, and label the second health and the corresponding historical operation parameters in the historical experimental data and the third simulated data to obtain the second labeled data.

[0119] In one embodiment, the generating unit 202 is specifically configured to: according to the actual measurement data, perform fitting optimization on the second model parameters of the first target empirical model to obtain the second target empirical model, generate fourth simulated data according to the second target empirical model, and according to the association mapping relationship between the capacity decay rate and the health, label the third health and the corresponding operation parameters in the fourth simulated data and the actual measurement data to obtain the third labeled data.

[0120] In one embodiment, the prediction unit 203 is specifically configured to: when the current health is greater than the preset health, determine that the target battery is in a healthy state; when the current health is less than or equal to the preset health, determine that the target battery is in an unhealthy state.

[0121] In one embodiment, the prediction unit 203 is specifically configured to: obtain multiple current health degrees in real time according to a preset period, characterize the health degree trend based on the multiple current health degrees, determine the total number of charge and discharge cycles when reaching a preset health degree, and determine the service life of the target battery according to the total number of charge and discharge cycles and the current number of charge and discharge cycles.

[0122] Regarding the device in the above embodiment, the specific manners in which each unit module performs operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0123] Figure 5 is a schematic diagram of an electronic device provided by the present application. As Figure 5 , the electronic device 50 may include at least one processor 501 and a memory 503 for storing instructions executable by the processor. Among them, the processor 501 is configured to execute the instructions in the memory 503 to implement the battery health state prediction method in the following embodiments.

[0124] In addition, the electronic device 50 may further include a communication bus 502, at least one communication interface 504, an input device 506, and an output device 505.

[0125] The processor 501 may be a central processing unit (CPU), a microprocessing unit, an ASIC, or one or more integrated circuits for controlling the execution of the program of the solution of the present application.

[0126] The communication bus 502 may include a path for transmitting information between the above components.

[0127] The communication interface 504 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0128] The input device 506 is used to receive input signals and the output device 505 is used to output signals.

[0129] The memory 503 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory can exist independently and be connected to the processing unit through a bus. The memory can also be integrated with the processing unit.

[0130] Among them, the memory 503 is used to store the instructions for executing the solution of this application and is controlled by the processor 501 to execute. The processor 501 is used to execute the instructions stored in the memory 503, thereby implementing the functions in the method of this application.

[0131] In a specific implementation, as an embodiment, the processor 501 can include one or more CPUs, such as Figure 5 CPU0 and CPU1 in

[0132] In a specific implementation, as an embodiment, the electronic device 50 can include multiple processors, such as Figure 5 the processor 501 and the processor 507 in

[0133] This electronic device, as shown in Figure 5 includes: a processor 501 and a memory 503 for storing executable instructions of the processor 501; among them, the processor 501 is configured to execute the executable instructions to implement the battery health state prediction method in any of the above possible implementation manners. And it can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0134] The embodiments of the present application further provide a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by the processor of the battery health state prediction device, the battery health state prediction device can execute the battery health state prediction method according to any of the above possible implementation manners, and can achieve the same technical effects. To avoid repetition, details are not described herein again.

[0135] The embodiments of the present application further provide a computer program product, including a computer program or instructions. The computer program or instructions are executed by the processor to perform the battery health state prediction method according to any of the above possible implementation manners, and can achieve the same technical effects. To avoid repetition, details are not described herein again.

[0136] After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily conceive of other embodiments of the present application. The present application is intended to cover any variations, uses, or adaptations of the present application, which follow the general principles of the present application and include known common knowledge or conventional technical means in the technical field not disclosed in the present application. The specification and embodiments are only regarded as exemplary, and the true scope and spirit of the present application are pointed out by the following claims.

[0137] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present application is only limited by the appended claims.

Claims

1. A method for predicting the state of health of a battery, characterized in that, The method includes: Determine the actual measurement data generated during the operation of the target battery before the current time; the actual measurement data includes actual operating parameters and actual health status; the actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual charge-discharge cycle count of the target battery; Optimize the model parameters of a preset model that has been trained based on the third tag data generated from the actual measurement data to obtain an optimized target model; the preset model is trained through a supervised learning training method based on first tag data and second tag data. The first tag data is generated by data tagging according to the test data of the target battery; the second tag data is generated by data tagging according to historical experimental data; the historical experimental data is data generated by historical batteries of the same type as the target battery that have been put into use; Input the current operating parameters of the target battery at the current time into the target model to obtain the current health status.

2. The method for predicting the battery health state according to claim 1, wherein Before determining the actual measurement data generated during the operation of the target battery before the current time, the method includes: Conduct a battery standardization test on the target battery to obtain the test data; the test data includes the test voltage, test current, test temperature, test operating time, test charge-discharge cycle count, and test health status of the target battery; According to the test data, fit and optimize the first model parameters of an empirical model constructed based on the capacity attenuation rate, temperature, and capacity under charge-discharge cycles to obtain a first target empirical model; the empirical model is constructed based on the Arrhenius formula for temperature change; Generate first simulation data according to the first target empirical model; According to the association mapping relationship between the capacity attenuation rate and the health status, tag the first health status and the corresponding test operating parameters in the first simulation data and the test data to obtain the first tag data.

3. The battery health state prediction method according to claim 1, characterized in that, Before determining the actual measurement data generated during the operation of the target battery before the current time, the method includes: Tag the second health status and the corresponding historical operating parameters in the historical experimental data to obtain the second tag data.

4. The method for predicting the battery health state according to claim 3, wherein The tagging of the second health status of the historical battery according to the historical experimental data includes: Determine the data interval of the historical experimental data; Determine that the data interval in the historical experimental data is less than a preset interval, and insert second simulation data into the historical experimental data using interpolation; Tag the second health status and the corresponding historical operating parameters in the historical experimental data and the second simulation data to obtain the second tag data.

5. The method for predicting the battery health state according to claim 4, wherein The method further includes: Determine that the data interval in the historical experimental data is greater than or equal to the preset interval, and insert third simulation data into the historical experimental data using extrapolation; Tag the second health status and the corresponding historical operating parameters in the historical experimental data and the third simulation data to obtain the second tag data.

6. The method for predicting the battery health state according to any one of claims 2 to 5, characterized in that, Generating the third tag data according to the actual measurement data includes: According to the actual measurement data, the second model parameters of the first target empirical model are fitted and optimized to obtain a second target empirical model; According to the second target empirical model, fourth simulation data is generated; According to the associated mapping relationship between the capacity attenuation rate and the health degree, the third health degree and the corresponding operating parameters in the fourth simulation data and the actual measurement data are labeled to obtain the third labeled data.

7. The method for predicting the state of health of a battery according to any one of claims 1 to 5, characterized in that, The method further includes: When the current health degree is greater than the preset health degree, it is determined that the target battery is in a healthy state; When the current health degree is less than or equal to the preset health degree, it is determined that the target battery is in an unhealthy state.

8. The method for predicting the battery health state according to claim 7, wherein The method further includes: According to a preset period, multiple current health degrees are obtained in real time; According to the trend of the health degree characterized by multiple current health degrees, the total number of charge and discharge cycles when reaching the preset health degree is determined; According to the total number of charge and discharge cycles and the current number of charge and discharge cycles, the service life of the target battery is determined.

9. A battery state of health prediction device, characterized in that, The device includes: The determination unit is configured to determine the actual measurement data generated during the operation of the target battery before the current time; the actual measurement data includes actual operating parameters and actual health degree; the actual operating parameters include the actual voltage, actual current, actual temperature, actual operating time, and actual number of charge and discharge cycles of the target battery; The generation unit is configured to optimize the model parameters of a preset model that has been trained based on the third labeled data generated according to the actual measurement data to obtain an optimized target model; the preset model is trained by a supervised learning training method based on first labeled data and second labeled data, the first labeled data is generated by data labeling according to the test data of the target battery; the second labeled data is generated by data labeling according to historical experimental data; the historical experimental data is data generated by a historical battery of the same type as the target battery and that has been put into use; The prediction unit is configured to input the current operating parameters of the target battery at the current time into the target model to obtain the current health degree.

10. A computer-readable storage medium, on which instructions are stored, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the battery health status prediction device, the battery health status prediction device can execute the battery health status prediction method according to any one of claims 1-8.

Images