Methods, devices, electronic equipment, and storage media for predicting the maximum usable capacity of batteries

Abstract

This application provides a method, apparatus, electronic device, and storage medium for predicting the maximum usable capacity of a battery. The method includes: acquiring characteristic data of the battery based on historical operating data; the characteristic data includes constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, capacity increment curve characteristics, and voltage difference characteristics; using a preset fitting function to perform temperature compensation and current compensation on the characteristic data to obtain compensated characteristic data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current, and battery capacity retention rate; inputting the compensated characteristic data into a trained machine learning model, and predicting the current maximum usable capacity of the battery through the trained machine learning model. This application embodiment can maintain prediction accuracy even in complex operating conditions, while also maintaining low prediction costs, thus simultaneously meeting the requirements of low cost and high accuracy.

Description

Technical Field

[0001] This application relates to the field of new energy, specifically to a method, apparatus, electronic device, and storage medium for predicting the maximum usable capacity of a battery. Background Technology

[0002] In the field of new energy, various batteries, including lithium batteries, are frequently used in new energy vehicles and various energy storage systems. However, during battery use, as the battery undergoes continuous charge-discharge cycles, its internal physicochemical properties undergo irreversible changes. This change leads to an irreversible decline in the battery's maximum usable capacity, further reducing its remaining lifespan and, in severe cases, affecting its safety performance. Therefore, predicting the maximum usable capacity of a battery is of great significance. However, current technologies for predicting the maximum usable capacity of batteries struggle to simultaneously meet the requirements of low cost and high accuracy under complex operating conditions. Summary of the Invention

[0003] One objective of this application is to provide a method, apparatus, electronic device, and storage medium for predicting the maximum usable capacity of a battery, which can maintain prediction accuracy even under complex operating conditions, while also maintaining low prediction costs, thus balancing the requirements of low cost and high accuracy.

[0004] According to one aspect of the embodiments of this application, a method for predicting the maximum usable capacity of a battery is disclosed, the method comprising:

[0005] Based on the battery's historical operating data, characteristic data of the battery are obtained; the characteristic data includes the battery's constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, capacity increment curve characteristics, and voltage difference characteristics.

[0006] Using a preset fitting function, temperature compensation and current compensation are performed on the feature data to obtain compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate.

[0007] The compensated feature data is input into the trained machine learning model, and the trained machine learning model is used to predict the current maximum usable capacity of the battery.

[0008] According to one aspect of the embodiments of this application, a battery maximum usable capacity prediction device is disclosed, the device comprising:

[0009] The feature data acquisition module is configured to acquire the feature data of the battery based on the battery's historical operating data; the feature data includes the battery's constant current charging time feature, constant voltage charging time feature, charging temperature rise feature, capacity increment curve feature, and voltage difference feature;

[0010] The dual compensation module is configured to use a preset fitting function to perform temperature compensation and current compensation on the feature data to obtain compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate.

[0011] The prediction module is configured to input the compensated feature data into a trained machine learning model and predict the current maximum available capacity of the battery through the trained machine learning model.

[0012] In an exemplary embodiment of this application, the historical operating data includes charge / discharge current data, charge / discharge voltage data, charge / discharge time data, and charge / discharge temperature data; the feature data acquisition module is configured as follows:

[0013] Based on the charging and discharging current data, the charging and discharging voltage data, the charging and discharging time data, and the charging and discharging temperature data, the constant current charging time feature, the constant voltage charging time feature, and the charging temperature rise feature are extracted.

[0014] The historical operating data is processed using the differential voltage method to extract a set of capacity increment curve features under various voltage differences; the set of capacity increment curve features includes the capacity increment curve features under each voltage difference.

[0015] Target capacity increment curve features and corresponding voltage difference features are selected from the set of capacity increment curve features.

[0016] In an exemplary embodiment of this application, the feature data acquisition module is configured as follows:

[0017] Calculate the entropy value of the characteristic data of the capacity increment curve under each voltage difference. The entropy value is used to describe the stability of the corresponding characteristic data of the capacity increment curve.

[0018] The capacity increment curve feature with the minimum entropy value is selected as the target capacity increment curve feature, and the voltage difference corresponding to the target capacity increment curve feature is used as the voltage difference feature.

[0019] In one exemplary embodiment of this application, the device is configured as follows:

[0020] The feature data is processed using an outlier detection algorithm to remove outliers from the feature data.

[0021] In one exemplary embodiment of this application, the device is configured as follows:

[0022] Each feature contained in the feature data is normalized.

[0023] In one exemplary embodiment of this application, the device is configured as follows:

[0024] Obtain a first function that describes the coupling relationship between charging temperature and battery capacity retention when charging at a constant current.

[0025] Obtain a second function that describes the coupling relationship between charging current and battery capacity retention when charging at a constant temperature.

[0026] The fitting function is constructed based on the product of the first function and the second function, and the mean of the first function and the second function.

[0027] In one exemplary embodiment of this application, the device is configured as follows:

[0028] The compensated feature data of the sample battery is input into the machine learning model to obtain the current maximum usable capacity of the sample battery output by the machine learning model; the compensated feature data of the sample battery is obtained by using the fitting function to perform temperature compensation and current compensation on the feature data of the sample battery.

[0029] A fitness function is constructed based on the deviation between the actual maximum usable capacity of the sample battery and the current maximum usable capacity of the sample battery.

[0030] The machine learning model is trained based on the iteration stopping condition constructed by the fitness function to obtain the trained machine learning model.

[0031] According to one aspect of the embodiments of this application, an electronic device is disclosed, comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device enables the battery maximum usable capacity prediction method provided in any of the above embodiments.

[0032] According to one aspect of the embodiments of this application, a computer program medium is disclosed, on which computer-readable instructions are stored, which, when executed by a computer's processor, cause the computer to perform the battery maximum usable capacity prediction method provided in any of the above embodiments.

[0033] According to one aspect of the embodiments of this application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the battery maximum usable capacity prediction method provided in the various embodiments described above.

[0034] In this embodiment, after obtaining the battery's feature data, a fitting function describing the coupling relationship between the battery's charging temperature, charging current, and battery capacity retention rate is used to perform temperature and current compensation on the feature data, thereby reducing the impact of charging temperature and charging current on the maximum usable capacity. The compensated feature data is then input into a trained machine learning model, reducing the influence of charging temperature and charging current during the model's prediction of the battery's current maximum usable capacity, thus ensuring the accuracy of the predicted maximum usable capacity. Even under complex operating conditions, prediction accuracy is maintained while keeping prediction costs low, thus simultaneously achieving the requirements of low cost and high accuracy.

[0035] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0036] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit this application. Attached Figure Description

[0037] The above and other objectives, features and advantages of this application will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0038] Figure 1 A flowchart of a method for predicting the maximum usable capacity of a battery according to an embodiment of this application is shown.

[0039] Figure 2 A schematic diagram illustrating the coupling relationship between the charging temperature, charging current, and battery capacity retention rate of a battery according to an embodiment of this application is shown.

[0040] Figure 3 A flowchart illustrating the prediction of battery health status according to one embodiment of this application is shown.

[0041] Figure 4 A flowchart illustrating the optimization of a random forest regression model according to one embodiment of this application is shown.

[0042] Figure 5A flowchart illustrating the training and testing of a random forest regression model according to one embodiment of this application is shown.

[0043] Figure 6 A block diagram of a battery maximum available capacity prediction device according to an embodiment of this application is shown.

[0044] Figure 7 A hardware diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0045] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided to make the description of this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The drawings are merely illustrative of this application and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0046] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced with one or more of the specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0047] Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0048] The following methods are commonly used in related technologies to predict the maximum usable capacity of a battery:

[0049] One method involves directly predicting the battery's maximum usable capacity based on data obtained from direct measurements by sensors, according to changes in data volume during the battery aging process. While this method is simple to calculate and low in cost, it cannot accurately predict the maximum usable capacity of a battery under complex operating conditions, resulting in significant errors in actual use under such conditions.

[0050] Second, an improved particle filter algorithm is developed using advanced mathematical methods. Based on this improved algorithm and the battery's internal model, a physical model is established to track the maximum usable capacity. This physical model is then used to predict the battery's maximum usable capacity. While this method offers high prediction accuracy, the difficulty of establishing the physical model increases with the complexity of operating conditions, leading to higher prediction costs under complex conditions.

[0051] As can be seen from the two methods provided by the above-mentioned related technologies, when predicting the maximum usable capacity of a battery, it is difficult to simultaneously meet the requirements of low cost and high accuracy in complex operating conditions.

[0052] In view of the above-mentioned defects in related technologies, this application provides a method for predicting the maximum usable capacity of a battery, which can simultaneously meet the requirements of low cost and high accuracy even in the face of complex operating conditions.

[0053] The battery maximum usable capacity prediction method provided in this application embodiment can be applied to a server or control terminal. It should be noted that the battery can be a battery in an energy storage device or a battery in an electrical device. The energy storage device or electrical device can send the battery's operating data to the server. Alternatively, the energy storage device or electrical device can send the battery's operating data to the control terminal. The server or control terminal executes the scheme provided in this embodiment to predict the battery's maximum usable capacity.

[0054] Figure 1 A flowchart illustrating the battery maximum usable capacity prediction method provided in this application is shown. See also... Figure 1 The methods provided in this application include:

[0055] Step S110: Based on the battery's historical operating data, obtain the battery's characteristic data; the characteristic data includes the battery's constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, capacity increment curve characteristics, and voltage difference characteristics.

[0056] In this embodiment, historical operating data refers to the battery's past charge and discharge data. In specific implementation, the battery is equipped with a battery management system (BMS), which enables the charging and discharging of the energy storage device's battery. An analog front-end chip can also be configured in the battery to collect data such as current and voltage. Additionally, a temperature sensor can be configured on the battery to collect its temperature. During the charging and discharging process, the BMS can aggregate all collected operating data and send it to a server or control terminal.

[0057] Understandably, the battery management system collects historical operating data corresponding to the battery and stores this data in a pre-set memory. The server or control terminal then communicates with the energy storage device or electrical appliance containing the battery to read this historical operating data.

[0058] As an example, based on the battery's historical operating data, the following battery characteristic data were calculated: constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, capacity increment curve characteristics, and voltage difference characteristics.

[0059] Among these features, constant current charging time characteristics primarily describe the time required to charge a battery to full capacity using a constant current. Constant voltage charging time characteristics primarily describe the time required to charge a battery to full capacity using a constant voltage. Charging temperature rise characteristics primarily describe the temperature rise of the battery during the charging process to full capacity. Capacity increment curve characteristics are often linked to voltage difference characteristics; the capacity increment curve characteristic under a specific voltage difference characteristic primarily describes the relationship between capacity increment and charging voltage under the voltage difference described by that voltage difference characteristic. Capacity increment curve characteristics can also be called IC curve characteristics.

[0060] Step S120: Using a preset fitting function, perform temperature compensation and current compensation on the feature data to obtain compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate.

[0061] Unlike technologies that use physical models to predict the maximum usable capacity of a battery, this application employs a machine learning model to predict the maximum usable capacity of the battery. Specifically, the battery's feature data is primarily used as input to the trained machine learning model, which then predicts and outputs the battery's maximum usable capacity based on this feature data.

[0062] However, in this embodiment, after obtaining the battery's feature data, the feature data is not directly input into the trained machine learning model. This is because the battery's maximum usable capacity is affected not only by usage time but also by charging temperature and charging current. Therefore, this embodiment also utilizes a preset fitting function to perform temperature compensation and current compensation on the feature data.

[0063] Figure 2 A schematic diagram illustrating the coupling relationship between battery charging temperature, charging current, and battery capacity retention rate in one embodiment is shown. Figure 2It is evident that battery capacity retention is related to both charging temperature and charging current. Battery capacity retention describes the ratio of the battery's capacity at full charge after a certain number of charge-discharge cycles under test conditions to its capacity at the time of manufacture. In other words, battery capacity retention is closely related to the battery's maximum usable capacity. Therefore, the maximum usable capacity of a battery is related to both charging temperature and charging current.

[0064] Therefore, after obtaining the battery's characteristic data, a preset fitting function is used to describe the coupling relationship between charging temperature, charging current, and battery capacity retention rate. Temperature compensation is applied to the characteristic data to reduce the impact of charging temperature on the battery's maximum usable capacity, and current compensation is applied to the characteristic data to reduce the impact of charging current on the battery's maximum usable capacity. Thus, the compensated characteristic data simultaneously reduces the impact of charging temperature and charging current on the battery's maximum usable capacity.

[0065] Step S130: Input the compensated feature data into the trained machine learning model, and predict the current maximum usable capacity of the battery through the trained machine learning model.

[0066] It should be noted that, generally, the maximum usable capacity is predicted after the battery has been in use for a period of time after leaving the factory. The maximum usable capacity after a period of use is usually lower than the maximum usable capacity at the time of manufacture, and the maximum usable capacity at the time of manufacture is usually referred to as the rated maximum usable capacity. To distinguish it from the rated maximum usable capacity, the maximum usable capacity to be predicted is referred to as the current maximum usable capacity in this application.

[0067] In this embodiment, the compensated feature data is input into the trained machine learning model, which then predicts the battery's current maximum usable capacity based on the compensated feature data. Since the compensated feature data simultaneously reduces the impact of charging temperature and charging current on the maximum usable capacity, the influence of charging temperature and charging current is reduced during the machine learning model's prediction of the battery's current maximum usable capacity, thus ensuring the accuracy of the predicted maximum usable capacity. Because the feature data processing is applicable to various operating conditions, this method of using feature data to drive the machine learning model for prediction maintains prediction accuracy even under complex operating conditions.

[0068] Furthermore, since the prediction using machine learning models does not require consideration of the complex internal physical model of the battery, the implementation difficulty of the embodiments of this application hardly increases with the complexity of the operating conditions. Therefore, even under complex operating conditions, the embodiments of this application can maintain a low prediction cost.

[0069] In summary, in this embodiment, after obtaining the battery's feature data, a fitting function describing the coupling relationship between the battery's charging temperature, charging current, and battery capacity retention rate is used to perform temperature and current compensation on the feature data, thereby reducing the impact of charging temperature and charging current on the maximum usable capacity. The compensated feature data is then input into a trained machine learning model, reducing the influence of charging temperature and charging current during the model's prediction of the battery's current maximum usable capacity, thus ensuring the accuracy of the predicted maximum usable capacity. Even under complex operating conditions, prediction accuracy is maintained while keeping prediction costs low, thus simultaneously achieving the requirements of low cost and high accuracy.

[0070] In one embodiment, the historical operating data includes charge / discharge current data, charge / discharge voltage data, charge / discharge time data, and charge / discharge temperature data. Accordingly, step S110 includes:

[0071] Based on the charging and discharging current data, charging and discharging voltage data, charging and discharging time data, and charging and discharging temperature data, the constant current charging time characteristics, constant voltage charging time characteristics, and charging temperature rise characteristics are extracted.

[0072] The differential voltage method is used to process historical operating data and extract a set of capacity increment curve features under various voltage differences; the set of capacity increment curve features includes the capacity increment curve features under each voltage difference.

[0073] The target capacity increment curve features and the corresponding voltage difference features are selected from the set of capacity increment curve feature data.

[0074] In this embodiment, the collected historical operating data includes charge / discharge current data, charge / discharge voltage data, charge / discharge time data, and charge / discharge temperature data. Specifically, the charge / discharge current data is mainly used to describe the current magnitude during the battery charge / discharge process. The charge / discharge voltage data is mainly used to describe the voltage magnitude during the battery charge / discharge process. The charge / discharge time data is mainly used to describe the timeline information during the battery charge / discharge process. The charge / discharge temperature data is mainly used to describe the temperature level during the battery charge / discharge process.

[0075] Furthermore, by combining charge / discharge current data and charge / discharge time data, we can extract the constant current charging time feature, which mainly describes the time taken to charge the battery to full capacity using a constant current. By combining charge / discharge voltage data and charge / discharge time data, we can extract the constant voltage charging time feature, which mainly describes the time taken to charge the battery to full capacity using a constant voltage. Finally, by combining charge / discharge temperature data and charge / discharge time data, we can extract the charging temperature rise feature, which mainly describes the temperature rise of the battery during the process of charging to full capacity.

[0076] For these historical operating data, the differential voltage method is used to process them, extracting capacity increment characteristic data that varies with charging voltage under various voltage differences. The capacity increment characteristic data can be denoted as dQ / dV, where Q is the charging quantity and V is the charging voltage. That is, the capacity increment characteristic data can be regarded as the derivative of the charging quantity Q with respect to the charging voltage V.

[0077] The characteristic data of capacity increment as a function of charging voltage under each voltage difference is the characteristic curve of capacity increment under the corresponding voltage difference. The characteristic curves of capacity increment corresponding to each of the multiple voltage differences are collected together to obtain the set of characteristic curve data of capacity increment.

[0078] Considering that not every capacity increment feature that varies with charging voltage under a voltage difference is suitable as input to a machine learning model, the set of capacity increment curve features is screened to select target capacity increment curve features that meet the model input requirements, and the voltage difference features corresponding to the target capacity increment curve features are determined.

[0079] It should be noted that the use of the differential voltage method in the process of extracting capacity increment curve features (i.e., IC curve features) avoids the high demand for computing power caused by the use of complex filtering algorithms in the traditional IC curve feature extraction process, thereby improving the extraction efficiency of IC curve features.

[0080] In one embodiment, selecting target capacity increment curve features and corresponding voltage difference features from a set of capacity increment curve feature data includes:

[0081] Calculate the entropy value of the characteristic data of the capacity increment curve under each voltage difference. The entropy value is used to describe the stability of the corresponding characteristic data of the capacity increment curve.

[0082] The capacity increment curve feature with the minimum entropy value is selected as the target capacity increment curve feature, and the voltage difference corresponding to the target capacity increment curve feature is used as the voltage difference feature.

[0083] In this embodiment, to ensure the stability of the feature data, the entropy value of the capacity increment curve feature data under each voltage difference is calculated. The smaller the entropy value, the more stable the corresponding capacity increment curve feature data. Conversely, the larger the entropy value, the less stable the corresponding capacity increment curve feature data.

[0084] Then, the capacity increment curve feature with the minimum entropy value is selected as the target capacity increment curve feature to be input into the machine learning model, and the voltage difference corresponding to the target capacity increment curve feature is used as the voltage difference feature to be input into the machine learning model.

[0085] In one embodiment, after the step of acquiring the battery's feature data, the method provided in this application further includes:

[0086] An outlier detection algorithm is used to process the feature data and remove outliers.

[0087] In this embodiment, it is considered that under normal circumstances, the feature data of the same battery usually exhibits a clustered distribution. If a feature data point deviates from the group, it indicates that the feature data is likely to be outlier. Therefore, to ensure the quality of the feature data input into the machine learning model, each feature data point is treated as a data point, and an outlier detection algorithm is used to process the feature data, detect outliers, and then remove the outliers.

[0088] In one embodiment, the LOF (Local Outlier Factor) algorithm is used to detect outliers in the feature data. Specifically, the following formula is used to detect outliers in the feature data:

[0089]

[0090] Where, N k (p) represents the k-distance neighborhood of point p; o is N k (p) A point within the neighborhood; lrd k (p) represents the local reachability density of point p; lrd k (o) represents the local reachability density of point o; LOF k (p) represents the k-th local outlier at point p. LOF k The greater the value of (p), the less density point p is compared to its surrounding points, and the more likely point p is to be an outlier; conversely, the greater the value of LOF, the less density point p is compared to its surrounding points. k The less than 1 the value of (p), the greater the density of point p is than the density of its surrounding points, and the more likely point p is to be a normal point.

[0091] In one embodiment, after the step of acquiring the battery's feature data, the method provided in this application further includes:

[0092] Each feature in the feature data is normalized separately.

[0093] In this embodiment, to ensure that the prediction accuracy is not reduced due to differences in data scale, each feature contained in the feature data is normalized. That is, the constant current charging time feature, constant voltage charging time feature, charging temperature rise feature, capacity increment curve feature, and voltage difference feature are normalized respectively, so that these feature data are processed under a unified data scale to ensure prediction accuracy.

[0094] In one embodiment, the Min_Max algorithm is used to normalize the feature data. Specifically, the feature data is normalized using the following formula:

[0095]

[0096] Where x represents the feature data to be normalized; x min x represents the smallest value in the feature data; max x represents the largest value in the feature data. norm This represents the normalized feature data.

[0097] In one embodiment, the method provided in this application further includes:

[0098] Obtain a first function that describes the coupling relationship between charging temperature and battery capacity retention when charging at a constant current.

[0099] Obtain a second function that describes the coupling relationship between charging current and battery capacity retention when charging at a constant temperature.

[0100] A fitting function is constructed based on the product of the first function and the second function, and the mean of the first function and the second function.

[0101] In this embodiment, a fitting function is constructed to describe the coupling relationship between the battery's charging temperature, charging current, and battery capacity retention rate by controlling variables.

[0102] Specifically, the charging current of the battery is kept constant, and then the battery capacity retention rate at different charging temperatures is obtained. Based on this, a first function is constructed to describe the coupling relationship between charging temperature and battery capacity retention rate.

[0103] Similarly, by keeping the battery charging temperature constant, the battery capacity retention rate under different charging currents is obtained, and then a second function is constructed to describe the coupling relationship between the charging current and the battery capacity retention rate.

[0104] Then, by multiplying and averaging the first and second functions, the first and second functions are coupled to construct a fitting function that describes the coupling relationship between the battery's charging temperature, charging current, and battery capacity retention rate.

[0105] As an example, by setting the charging current multiplier to a constant 0.5, we construct a first function D(x,0.5) to describe the coupling relationship between charging temperature and battery capacity retention, where x represents the charging temperature:

[0106] D(x,0.5)=-9.191e-05·x 3 -0.002755·x 2 +0.6307·x+84.37

[0107] By setting the charging temperature factor to a constant value of 5, a second function D(5,y) is constructed to describe the coupling relationship between the charging current and the battery capacity retention rate, where y represents the charging current:

[0108] D(5,y)=-6.141·y 3 +22.83·y 2 -27.49·y+100.8

[0109] Then, taking the average of these two functions, we get:

[0110]

[0111] Then, by combining the product and mean of these two functions, a fitting function D(x,y) is constructed to describe the coupling relationship between the battery's charging temperature, charging current, and battery capacity retention rate:

[0112]

[0113] It is understood that all the values ​​involved in the above examples are calibrated based on actual measurements of battery cells made of the same material. In practical implementation, multiple measurements can be performed on battery cells made of different materials, and the relevant experimental data can be summarized into a table for easy reference and selection. Here, since those skilled in the art can measure and calibrate different parameters based on the different battery cells actually selected, and then construct the first function and the second function, as well as couple the first function and the second function, the values ​​calibrated based on actual measurements in the above examples will not be elaborated here.

[0114] In one embodiment, the method provided in this application further includes:

[0115] The compensated feature data of the sample battery is input into the machine learning model to obtain the current maximum usable capacity of the sample battery output by the machine learning model. The compensated feature data of the sample battery is obtained by using a fitting function to perform temperature compensation and current compensation on the feature data of the sample battery.

[0116] The fitness function is constructed based on the deviation between the true maximum usable capacity of the sample battery and the current maximum usable capacity of the sample battery.

[0117] Based on the iteration stopping condition constructed using the fitness function, the machine learning model is trained to obtain the trained machine learning model.

[0118] In this embodiment, the sample battery refers to the battery used when training the machine learning model.

[0119] After obtaining the characteristic data of the sample battery, namely the constant current charging time characteristic, constant voltage charging time characteristic, charging temperature rise characteristic, capacity increment curve characteristic, and voltage difference characteristic, the characteristic data of the sample battery are compensated for temperature and current using a fitting function that describes the coupling relationship between charging temperature, charging current and battery capacity retention rate, to obtain the compensated characteristic data of the sample battery.

[0120] Then, the compensated feature data of the sample battery is input into the machine learning model to obtain the current maximum usable capacity of the sample battery as output by the machine learning model.

[0121] Before a machine learning model is fully trained, its output of the current maximum usable capacity of a sample battery is typically inaccurate. To ensure the machine learning model outputs accurate results, the deviation between the true maximum usable capacity of the sample battery and the current maximum usable capacity output by the machine learning model is calculated. Here, the true maximum usable capacity of the sample battery is known.

[0122] Then, based on the calculated bias, a fitness function is constructed to measure the model's performance. Based on this fitness function, an iteration stopping condition for the machine learning model is then constructed. The machine learning model is then iteratively trained until the iteration stopping condition is met, resulting in the trained machine learning model.

[0123] In one embodiment, the machine learning model is a random forest regression model constructed using the random forest regression algorithm. In this case, iterative training of the random forest regression model mainly involves iteratively optimizing two model parameters: the number of random forest trees and the number of random forest nodes.

[0124] Preferably, the Moth to a Flame algorithm is used to iteratively optimize the random forest regression model. Specifically, after constructing the fitness function, the stopping condition for the Moth to a Flame algorithm is constructed based on the fitness function. Then, the Moth to a Flame algorithm is used to iteratively adjust the random forest tree quantity parameter and random forest node quantity parameter of the random forest regression model to train and optimize the random forest regression model.

[0125] In one embodiment, a machine learning model is constructed using the random forest regression algorithm to obtain the random forest regression model to be trained. Then, the compensated feature data of the sample battery is input into the random forest regression model to obtain the current maximum usable capacity of the sample battery output by the random forest regression model. Subsequently, the current health state (SOH) of the sample battery corresponding to the prediction result of the random forest regression model is calculated according to the following formula. m :

[0126] SOH m = Current maximum usable capacity of the sample battery / Rated maximum usable capacity of the sample battery

[0127] Then, the current health status (SOH) of the sample batteries corresponding to the prediction results of the random forest regression model is determined. m Compared with the true health status (SOH) of the sample battery z By taking the difference and the absolute value, the fitness function δ=|SOH is constructed. m -SOH z Then, based on the fitness function, the iterative stopping condition of the Moth to a Flame algorithm is constructed. The Moth to a Flame algorithm is then used to iteratively adjust the random forest tree quantity parameter and random forest node quantity parameter of the random forest regression model in order to train and optimize the random forest regression model.

[0128] Among them, the true health status (SOH) of the sample batteries z The actual maximum usable capacity of the sample battery is calculated using the following formula:

[0129] SOH z = Actual maximum usable capacity of the sample battery / Rated maximum usable capacity of the sample battery

[0130] It should be noted that since the battery's state of health (SOH) = (current maximum usable capacity of the battery / rated maximum usable capacity of the battery) * 100%, and the rated maximum usable capacity of the battery is usually fixed and known at the time of manufacture, the processing logic for predicting the maximum usable capacity of the battery according to the method provided in this application can be directly applied to predicting the battery's state of health.

[0131] Specifically, Figure 3 A flowchart for predicting the health status of a battery is shown in one embodiment of this application.

[0132] See Figure 3 In one embodiment, data cleaning is performed on the historical operating data of the sample battery to remove duplicate data and data containing invalid values.

[0133] Then, from the cleaned historical operating data, the following characteristic data of the sample battery were extracted: constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, IC curve characteristics, and voltage difference characteristics. Specifically, the IC curve characteristics and voltage difference characteristics were extracted using the differential voltage method to obtain IC curve characteristic data under various voltage differences. Further, after extracting the IC curve characteristics under various voltage differences, the entropy value of the IC curve characteristic data under each voltage difference was calculated, and then the one with the smallest entropy value was selected.

[0134] Then, the feature data of the sample data undergoes dual compensation, outlier removal, and normalization. Specifically, dual compensation refers to using a pre-defined fitting function to describe the coupling relationship between charging temperature, charging current, and battery capacity retention rate to perform temperature and current compensation on the feature data of the sample battery. Outlier removal refers to using the LOF algorithm to detect outliers in the feature data of the sample battery and then removing them. Normalization refers to using the Min-Max algorithm to normalize the feature data of the sample battery.

[0135] After completing double compensation processing, outlier removal, and normalization, the feature data of the sample data is used to optimize and train the random forest regression model. Specifically, the feature data of the sample battery is denoted as x, resulting in the input vector X = {x1, x2, ..., xm} to be input into the machine learning model, where m represents the number of training samples. Then, the input vector X is transformed into the input sequence format required by the random forest regression model and input into the model, yielding the predicted result Y = {C1, C2, ..., Cm}, where C represents the model's predicted current health state. The random forest regression model is then optimized based on the deviation between the actual health state and the current health state of the sample battery until it can accurately predict the health state of the sample battery, completing the model training. The actual health state of the sample battery is calculated as follows: complete charging current data for any cycle is extracted from the feature data of the sample battery, and then the actual maximum usable capacity of the sample battery is calculated using the ampere-hour integration method, thereby calculating the actual health state of the sample battery.

[0136] After optimizing and training the random forest regression model, the trained random forest regression model is put into use to predict the health status of batteries under various operating conditions.

[0137] Figure 4 A flowchart illustrating the optimization of a random forest regression model in one embodiment of this application is shown.

[0138] See Figure 4 In one embodiment, the moth-to-a-flame algorithm is used to optimize the random forest regression model.

[0139] Specifically, step S1: Perform parameter initialization. Here, parameter initialization is to determine parameters such as the size of the moth population, the dimension of variables, and the maximum number of iterations.

[0140] Then, based on the initialized parameters, step S2 is executed: the positions of individual moths are calculated within the desired range, and moth positions are randomly generated. In the specific implementation, an initial moth matrix can also be established simultaneously. During each iteration, step S3 is executed: based on the fitness value of the individual moths, the moths are driven to search for and update flames. That is, flames with better fitness are searched, and flames with poorer fitness are discarded.

[0141] As the process gradually reduces the amount of flame, when only one flame remains, the moths will flock to that flame and converge on it.

[0142] Based on the constructed fitness function, the fitness values ​​of individual moths are calculated and updated. The moths are then grouped and sorted according to their fitness values, and this sorting information is transmitted to the flame, updating the flame information in real time. Because moths are phototactic, if they lock onto a flame, they will use a spiral flight pattern within a designated area to capture it. The endpoint is the position of the flame. The fitness values ​​of the individual moths are updated again. Using the updated results, the fitness values ​​are compared with those of the flame, and the best fitness value is assigned to the flame.

[0143] After driving the moth to search for and update the flame, proceed to step S4: determine whether the updated flame meets the termination condition.

[0144] If the termination condition is met, proceed to step S8: obtain and output the random forest tree quantity parameter and the random forest node quantity parameter.

[0145] If the termination condition is not met, proceed to step S5: update the flame quantity and position, and iterate again to drive the moth to search for and update the flame. Simultaneously, proceed to step S6: increment the iteration count K by one until the termination condition is met. Then proceed to step S7: determine if the iteration count K has reached the maximum iteration count K. max If so, proceed to step S8: obtain and output the parameters for the number of random forest trees and the number of random forest nodes.

[0146] Figure 5The flowchart illustrating the training and testing of a random forest regression model in one embodiment of this application is shown.

[0147] See Figure 5 In one embodiment, four sample batteries are provided. After performing double compensation processing, outlier removal processing, and normalization processing on the feature data of the sample batteries, the feature data of the first sample battery is used as the training set, and the feature data of the other three sample batteries is used as the test set.

[0148] After training the random forest regression model using the training set, the test data from the test set is input into the random forest regression model to obtain the prediction results corresponding to the test set. Then, the root mean square error (RMSE) formula is used to evaluate the test results of the random forest regression model. The RMSE formula is as follows:

[0149]

[0150] Where RMSE represents the root mean square error; m represents the length of the test data; k represents the cycle number of the test data; y k This indicates the true health status of the test data; This represents the current health status predicted by the model.

[0151] In one embodiment, four sample cells are provided. The feature data of the first sample cell is used as the training set to train the random forest regression model, and the feature data of the other three sample cells are used as the test set to test the random forest regression model.

[0152] To verify the impact of dual compensation on model performance, two training and testing methods were set up. In the first method, temperature and current compensation were applied to the feature data of the four sample batteries before training and testing the random forest regression model. In the second method, temperature and current compensation were not applied to the feature data of the four sample batteries, and the random forest regression model was directly trained and tested.

[0153] Table 1 below shows the prediction error of the model obtained by training with double compensation in the first method, and Table 2 below shows the prediction error of the model obtained by training directly without double compensation in the second method.

[0154] Table 1. Prediction errors of the model obtained by performing double compensation followed by training

[0155] 2 1.01 3 0.88 4 0.63

[0156] Table 2. Prediction errors of models trained directly without double compensation

[0157] 2 1.94 3 2.03 4 2.16

[0158] Figure 6 A block diagram of a battery maximum available capacity prediction device according to an embodiment of this application is shown, the device comprising:

[0159] The feature data acquisition module 210 is configured to acquire the feature data of the battery based on the battery's historical operating data; the feature data includes the battery's constant current charging time feature, constant voltage charging time feature, charging temperature rise feature, capacity increment curve feature, and voltage difference feature;

[0160] The dual compensation module 220 is configured to use a preset fitting function to perform temperature compensation and current compensation on the feature data to obtain compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate.

[0161] The prediction module 230 is configured to input the compensated feature data into a trained machine learning model and predict the current maximum available capacity of the battery through the trained machine learning model.

[0162] In an exemplary embodiment of this application, the historical operating data includes charge / discharge current data, charge / discharge voltage data, charge / discharge time data, and charge / discharge temperature data; the feature data acquisition module is configured as follows:

[0163] Based on the charging and discharging current data, the charging and discharging voltage data, the charging and discharging time data, and the charging and discharging temperature data, the constant current charging time feature, the constant voltage charging time feature, and the charging temperature rise feature are extracted.

[0164] The historical operating data is processed using the differential voltage method to extract a set of capacity increment curve features under various voltage differences; the set of capacity increment curve features includes the capacity increment curve features under each voltage difference.

[0165] Target capacity increment curve features and corresponding voltage difference features are selected from the set of capacity increment curve features.

[0166] In an exemplary embodiment of this application, the feature data acquisition module is configured as follows:

[0167] Calculate the entropy value of the characteristic data of the capacity increment curve under each voltage difference. The entropy value is used to describe the stability of the corresponding characteristic data of the capacity increment curve.

[0168] The capacity increment curve feature with the minimum entropy value is selected as the target capacity increment curve feature, and the voltage difference corresponding to the target capacity increment curve feature is used as the voltage difference feature.

[0169] In one exemplary embodiment of this application, the device is configured as follows:

[0170] The feature data is processed using an outlier detection algorithm to remove outliers from the feature data.

[0171] In one exemplary embodiment of this application, the device is configured as follows:

[0172] Each feature contained in the feature data is normalized.

[0173] In one exemplary embodiment of this application, the device is configured as follows:

[0174] Obtain a first function that describes the coupling relationship between charging temperature and battery capacity retention when charging at a constant current.

[0175] Obtain a second function that describes the coupling relationship between charging current and battery capacity retention when charging at a constant temperature.

[0176] The fitting function is constructed based on the product of the first function and the second function, and the mean of the first function and the second function.

[0177] In one exemplary embodiment of this application, the device is configured as follows:

[0178] The compensated feature data of the sample battery is input into the machine learning model to obtain the current maximum usable capacity of the sample battery output by the machine learning model; the compensated feature data of the sample battery is obtained by using the fitting function to perform temperature compensation and current compensation on the feature data of the sample battery.

[0179] A fitness function is constructed based on the deviation between the actual maximum usable capacity of the sample battery and the current maximum usable capacity of the sample battery.

[0180] The machine learning model is trained based on the iteration stopping condition constructed by the fitness function to obtain the trained machine learning model.

[0181] The following is for reference. Figure 7 To describe the electronic device 30 according to an embodiment of this application. Figure 7 The electronic device 30 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0182] like Figure 7As shown, the electronic device 30 is presented in the form of a general-purpose computing device. The components of the electronic device 30 may include, but are not limited to: at least one processing unit 310, at least one storage unit 320, and a bus 330 connecting different system components (including storage unit 320 and processing unit 310).

[0183] The storage unit stores program code that can be executed by the processing unit 310, causing the processing unit 310 to perform the steps described in the exemplary method description section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 310 can perform actions such as... Figure 1 The steps shown are as follows.

[0184] Storage unit 320 may include readable media in the form of volatile storage units, such as random access memory (RAM) 3201 and / or cache memory 3202, and may further include read-only memory (ROM) 3203.

[0185] Storage unit 320 may also include a program / utility 3204 having a set (at least one) program module 3205, such program module 3205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0186] Bus 330 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0187] Electronic device 30 can also communicate with one or more external devices 400 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 30, and / or with any device that enables electronic device 30 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 350. Input / output (I / O) interface 350 is connected to display unit 340. Furthermore, electronic device 30 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 360. As shown, network adapter 360 communicates with other modules of electronic device 30 via bus 330. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 30, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0188] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the method according to the embodiments of this application.

[0189] In an exemplary embodiment of this application, a computer-readable storage medium is also provided, on which computer-readable instructions are stored, which, when executed by a computer's processor, cause the computer to perform the methods described in the above method embodiments.

[0190] According to one embodiment of this application, a program product for implementing the methods in the above-described method embodiments is also provided. This product may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of this invention is not limited thereto. In this document, a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0191] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0192] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0193] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0194] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as JAVA and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0195] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0196] Furthermore, although the steps of the method in this application are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0197] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the method according to the embodiments of this application.

[0198] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the appended claims.

Claims

1. A method for predicting the maximum usable capacity of a battery, characterized in that, The method includes: Based on the battery's historical operating data, characteristic data of the battery are obtained; the characteristic data includes the battery's constant current charging time characteristics, constant voltage charging time characteristics, charging temperature rise characteristics, capacity increment curve characteristics, and voltage difference characteristics. Using a preset fitting function, temperature compensation and current compensation are applied to the feature data to reduce the impact of charging temperature and charging current on the maximum usable capacity of the battery, resulting in compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate. The compensated feature data is input into the trained machine learning model, and the trained machine learning model is used to predict the current maximum usable capacity of the battery. The method further includes: Obtain a first function that describes the coupling relationship between charging temperature and battery capacity retention when charging at a constant current. Obtain a second function that describes the coupling relationship between charging current and battery capacity retention when charging at a constant temperature. The fitting function is constructed based on the product of the first function and the second function, and the mean of the first function and the second function. The fitting function is obtained based on the quotient of the product and the mean.

2. The method according to claim 1, characterized in that, The historical operating data includes charging and discharging current data, charging and discharging voltage data, charging and discharging time data, and charging and discharging temperature data. The characteristic data of the battery obtained based on the battery's historical operating data includes: Based on the charging and discharging current data, the charging and discharging voltage data, the charging and discharging time data, and the charging and discharging temperature data, the constant current charging time feature, the constant voltage charging time feature, and the charging temperature rise feature are extracted. The historical operating data is processed using the differential voltage method to extract a set of capacity increment curve features under various voltage differences; the set of capacity increment curve features includes the capacity increment curve features under each voltage difference. Target capacity increment curve features and corresponding voltage difference features are selected from the set of capacity increment curve features.

3. The method according to claim 2, characterized in that, The step of filtering out the target capacity increment curve feature and the voltage difference feature corresponding to the target capacity increment curve feature from the set of capacity increment curve feature data includes: Calculate the entropy value of the characteristic data of the capacity increment curve under each voltage difference. The entropy value is used to describe the stability of the corresponding characteristic data of the capacity increment curve. The capacity increment curve feature with the minimum entropy value is selected as the target capacity increment curve feature, and the voltage difference corresponding to the target capacity increment curve feature is used as the voltage difference feature.

4. The method according to claim 1, characterized in that, After the step of acquiring the feature data of the battery, the method further includes: The feature data is processed using an outlier detection algorithm to remove outliers from the feature data.

5. The method according to claim 1, characterized in that, After the step of acquiring the feature data of the battery, the method further includes: Each feature contained in the feature data is normalized.

6. The method according to claim 1, characterized in that, The method further includes: The compensated feature data of the sample battery is input into the machine learning model to obtain the current maximum usable capacity of the sample battery output by the machine learning model; the compensated feature data of the sample battery is obtained by using the fitting function to perform temperature compensation and current compensation on the feature data of the sample battery. A fitness function is constructed based on the deviation between the actual maximum usable capacity of the sample battery and the current maximum usable capacity of the sample battery. The machine learning model is trained based on the iteration stopping condition constructed by the fitness function to obtain the trained machine learning model.

7. A device for predicting the maximum usable capacity of a battery, characterized in that, The device includes: The feature data acquisition module is configured to acquire the feature data of the battery based on the battery's historical operating data; the feature data includes the battery's constant current charging time feature, constant voltage charging time feature, charging temperature rise feature, capacity increment curve feature, and voltage difference feature; The dual compensation module is configured to use a preset fitting function to perform temperature compensation and current compensation on the feature data to reduce the impact of charging temperature and charging current on the maximum usable capacity of the battery, and obtain compensated feature data; wherein, the preset fitting function is used to describe the coupling relationship between charging temperature, charging current and battery capacity retention rate. The prediction module is configured to input the compensated feature data into a trained machine learning model and predict the current maximum usable capacity of the battery through the trained machine learning model. The device is further configured to: acquire a first function describing the coupling relationship between charging temperature and battery capacity retention rate when charging at a constant current; acquire a second function describing the coupling relationship between charging current and battery capacity retention rate when charging at a constant temperature; and construct the fitting function based on the product of the first function and the second function, and the mean of the first function and the second function; wherein the fitting function is obtained based on the quotient of the product and the mean.

8. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores computer-readable instructions that, when executed by the processor of a computer, cause the computer to perform the method described in any one of claims 1 to 6.

Images