Method, device and system for predicting residual charging time of battery, medium and equipment

Abstract

The invention provides a battery residual charging time prediction method, device and system, a medium and equipment, and is applied to the technical field of automotive electronics. According to the method, target feature data of a power battery pack of a target vehicle are obtained, and the target feature data are target feature data obtained through feature screening and related to the remaining charging time of the power battery pack; and inputting the target feature data into a pre-trained residual charging time prediction model to obtain the residual charging time of the power battery pack output by the residual charging time prediction model. According to the method, the target feature data related to the battery residual charging time is obtained through feature screening and is input into the trained residual charging time prediction model, so that the model can more accurately understand and capture key factors influencing the charging time, and accurate prediction of the battery residual charging time is realized.

Description

Technical Field

[0001] The present invention relates to the field of automotive electronics technology, and in particular to a method, device, system, medium and equipment for predicting the remaining charging time of a battery. Background Art

[0002] With growing attention to environmental protection and sustainable development, electric vehicles are gaining widespread attention for their clean and renewable energy. Under current DC charging standards, vehicle controllers are required to output the remaining charge time of the power battery pack to help drivers and commercial fleets optimize their travel schedules and reduce charging costs.

[0003] However, current methods for predicting remaining charging time mainly rely on charge rate tables provided by battery manufacturers, without considering factors such as battery state of health (SOH) aging and temperature changes, resulting in reduced prediction accuracy.

[0004] Therefore, how to improve the prediction accuracy of the remaining charging time of the battery has become a technical problem that technicians in this field urgently need to solve. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method, device, system, medium and equipment for predicting the remaining charging time of a battery that overcomes the above problems or at least partially solves the above problems. The technical solution is as follows:

[0006] A method for predicting remaining battery charging time, comprising:

[0007] Obtaining target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to a remaining charging time of the power battery pack obtained through feature screening;

[0008] The target feature data is input into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

[0009] Optionally, the feature screening process includes:

[0010] Obtaining a pending feature set of the power battery pack and a corresponding actual remaining charging time, wherein the pending feature set includes a plurality of charging features of the power battery pack;

[0011] Analyzing the correlation between each charging feature and the actual remaining charging time and the redundancy between each charging feature in the pending feature set, and sorting the charging features from largest to smallest according to their comprehensive correlation to obtain a valid feature set;

[0012] The remaining charging time prediction model is trained using the effective feature set, and a plurality of target features related to the remaining charging time of the power battery pack are screened out from the effective feature set.

[0013] Optionally, analyzing the correlation between each charging feature and the actual remaining charging time and the redundancy between each charging feature in the undetermined feature set, and sorting the charging features from largest to smallest according to their comprehensive correlation, to obtain a valid feature set, includes:

[0014] respectively calculating the mutual information between each of the charging characteristics and the actual remaining charging time;

[0015] Transferring the charging feature with the largest mutual information from the pending feature set to the valid feature set;

[0016] Calculating comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, and between each charging feature in the valid feature set;

[0017] Transferring the charging feature with the largest comprehensive mutual information from the pending feature set to the valid feature set;

[0018] Determine whether the pending feature set is an empty set. If so, perform the step of calculating the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, and each charging feature in the valid feature set. If not, obtain the valid feature set sorted from large to small according to the comprehensive correlation.

[0019] Optionally, the using the effective feature set to perform model training on the remaining charging time prediction model, and screening out multiple target features related to the remaining charging time of the power battery pack from the effective feature set, includes:

[0020] Dividing the effective feature set into a plurality of hierarchically increasing subsets, each subset being an extension of the previous subset;

[0021] Inputting each subset into the remaining charging time prediction model in turn to obtain the predicted remaining charging time output by the remaining charging time prediction model;

[0022] The charging feature in the subset with the smallest error between the predicted remaining charging time and the actual remaining charging time is confirmed as a target feature related to the remaining charging time of the power battery pack.

[0023] Optionally, the charging characteristics include the whole pack SOC, whole pack SOH, battery pack voltage, single cell maximum voltage, single cell minimum voltage, ambient temperature, single cell maximum temperature, single cell minimum temperature and charging loop current characteristics of the power battery pack during the charging process.

[0024] Optionally, the target feature data is data obtained by cleaning the original signal of the target feature of the power battery pack.

[0025] A battery remaining charging time prediction device includes: a target feature data obtaining unit and a remaining charging time obtaining unit,

[0026] The target feature data obtaining unit is configured to obtain target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening;

[0027] The remaining charging time obtaining unit is used to input the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

[0028] A battery remaining charging time prediction system includes: a cloud, a vehicle terminal, and a battery management system, wherein the cloud and the vehicle terminal are connected via wireless transmission, and the vehicle terminal and the battery management system are connected via a CAN bus;

[0029] The vehicle terminal obtains original feature data of the vehicle's power battery pack from the battery management system, wherein the original feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening;

[0030] The vehicle terminal transmits the original feature data to the cloud;

[0031] After receiving the original feature data transmitted by the vehicle terminal, the cloud performs data cleaning on the original feature data to obtain target feature data;

[0032] The cloud inputs the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model;

[0033] The cloud feeds back the remaining charging time to the vehicle terminal;

[0034] The vehicle terminal transmits the remaining charging time to the battery management system.

[0035] A computer-readable storage medium stores a program, which implements the battery remaining charging time prediction method when the program is executed by a processor.

[0036] An electronic device includes at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other via the bus; the processor is used to call program instructions in the memory to execute the battery remaining charging time prediction method.

[0037] By means of the above technical solution, the present invention provides a method, apparatus, system, medium, and device for predicting the remaining charge time of a battery, which obtains target feature data of the power battery pack of a target vehicle, wherein the target feature data is target feature data related to the remaining charge time of the power battery pack obtained through feature screening; the target feature data is input into a pre-trained remaining charge time prediction model to obtain the remaining charge time of the power battery pack output by the remaining charge time prediction model. The present invention obtains target feature data related to the remaining charge time of the battery through feature screening and inputs it into the trained remaining charge time prediction model, which enables the model to more accurately understand and capture the key factors affecting the charging time, thereby achieving an accurate prediction of the remaining charge time of the battery.

[0038] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0040] Figure 1 A schematic diagram showing a flow chart of an implementation of a method for predicting remaining battery charging time provided by an embodiment of the present invention;

[0041] Figure 2 A schematic diagram showing a process flow of an implementation method of feature screening provided by an embodiment of the present invention;

[0042] Figure 3 A schematic diagram showing a process of another implementation method of feature screening provided by an embodiment of the present invention;

[0043] Figure 4A schematic diagram showing the architecture of a battery remaining charging time prediction system provided by an embodiment of the present invention is shown;

[0044] Figure 5 A schematic diagram showing the structure of a battery remaining charging time prediction provided by an embodiment of the present invention is shown;

[0045] Figure 6 A schematic structural diagram of an electronic device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0046] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0047] With increasing attention to environmental protection and sustainable development, electric vehicles (EVs) have garnered widespread attention due to their clean and renewable energy advantages. DC charging is the most commonly used power supply for EVs. Current DC charging standards require the vehicle controller to output the remaining charge time. This refers to the time required for an EV to charge from its current state of charge (SOC) to its target SOC. Accurately understanding the remaining charge time not only helps drivers plan their future trips but also enables commercial fleets to prioritize charging needs while sizing charging times, thereby reducing charging costs and achieving optimal charging scheduling. Therefore, accurately predicting the remaining charge time is crucial.

[0048] Currently, remaining charge time prediction methods primarily rely on a charge rate table provided by battery manufacturers. This table takes the battery SOC and temperature as input and outputs the current maximum allowable charge rate. By looking up the table, the vehicle can obtain the maximum charge rate for a specific battery SOC and temperature. This can be used to calculate the remaining charge time offline for different SOC and temperature levels. The specific method is to divide the battery's SOC to be charged by the charge rate, and then accumulate the total remaining charge time.

[0049] However, existing prediction methods have some limitations. First, the charge rate table is based on test data from battery manufacturers under standard test conditions. It does not take into account the aging of the battery's State of Health (SOH) caused by static and cyclic charge and discharge during actual use, which may lead to a decrease in the prediction accuracy of aged batteries. Second, existing methods only consider the current SOC and temperature of the battery pack, and obtain the remaining charge time by accumulating the segmented charging time, assuming that the battery temperature remains unchanged during the charging process. However, in actual use, the battery temperature will change due to factors such as internal resistance heating, heat exchange in the thermal management system, and environmental heat convection. Predicting the remaining charge time based solely on the current temperature may cause errors.

[0050] Furthermore, existing technologies primarily predict the remaining charging time on the vehicle side. Considering the various factors that influence the remaining battery charging time will increase the computational burden on the vehicle-side chip. Centralized data processing in the cloud based on the intelligent Internet of Vehicles can provide new options for cost savings and efficient computing. Finally, while factors such as the battery pack's SOC, temperature, ambient temperature, and battery health status have an impact on prediction accuracy, not all factors are highly correlated with the remaining charging time at the data level. The input of irrelevant factors may reduce prediction accuracy, while correlations between factors may also lead to information redundancy and waste modeling computing power. Therefore, effectively selecting and integrating these influencing factors is key to improving the accuracy of remaining charging time predictions.

[0051] Based on this, the present invention provides a method for predicting the remaining charging time of a battery. First, by performing feature screening on the target vehicle's power battery pack and extracting feature data that is highly correlated with the remaining charging time, redundant information can be effectively reduced and the learning efficiency of the model can be improved. Secondly, these screened feature data are input into a pre-trained remaining charging time prediction model, allowing the model to more accurately understand and capture the key factors affecting charging time. It can be seen that this process not only optimizes data utilization efficiency, but also significantly improves the accuracy of the prediction results, thereby ensuring that the prediction model can better reflect the actual charging status of the battery in practical applications and achieve accurate prediction of the remaining charging time of the battery.

[0052] like Figure 1 As shown, a flowchart of an implementation of a method for predicting remaining battery charging time provided by an embodiment of the present invention is provided. The method may include:

[0053] S100: Obtain target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening.

[0054] The target vehicle is an electric or hybrid vehicle that requires charging. The power battery pack is the battery component in the target vehicle that stores electrical energy and is responsible for powering the electric vehicle. The power battery pack typically consists of multiple battery cells that store and release electrical energy to drive the vehicle.

[0055] Target feature data refers to the target feature data obtained through feature screening, which reflects the key features that affect the remaining charge time of the power battery pack. Feature screening is used to select features that are highly correlated with the remaining charge time from a large amount of power battery pack data. This helps remove redundant or irrelevant information and improves the efficiency and accuracy of the remaining charge time prediction model.

[0056] Among them, the remaining charging time refers to the remaining time expected to complete charging based on the current battery status of the power battery pack.

[0057] Specifically, embodiments of the present invention pre-collect data on the target vehicle's power battery pack, collecting various parameters that may affect the remaining charge time. Subsequently, through feature screening techniques, key features highly correlated with the remaining charge time are extracted from this data. These key features are referred to as target features. Ultimately, the target feature data of the target vehicle's power battery pack is collected online for subsequent remaining charge time prediction, accurately predicting the remaining charge time of the power battery pack and improving charging management efficiency.

[0058] Optionally, the target characteristic data is data obtained by cleaning the original signal of the target characteristic of the power battery pack.

[0059] Specifically, embodiments of the present invention can first apply the Three Sigma Rule to identify and remove outliers from the original signal of the target characteristic of the power battery pack. For outliers, linear extrapolation or interpolation methods are used to replace them to ensure data continuity and reliability. Next, the signal sequence after outlier removal is filtered and denoised using the moving average method, using a moving window of 5 sampling points to smooth the signal and reduce the impact of noise. Finally, the denoised data is normalized using the Z-score method to eliminate dimensionality effects and avoid weight bias, allowing the data to be compared and analyzed on a unified scale.

[0060] The Z-score standardization process is as follows:

[0061]

[0062] in, is the signal sequence of the original signal after denoising, is the average value of the signal sequence of the original signal after denoising, is the standard deviation of the signal sequence of the original signal after denoising, It is the Z score of the signal sequence of the original signal after denoising, that is, the target feature data used to reflect the degree of deviation of the data point from the mean.

[0063] The embodiment of the present invention ensures the accuracy and validity of the target feature data by performing data cleaning on the original signal of the target feature of the power battery pack, providing a reliable basis for subsequent analysis.

[0064] S110 : Input the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

[0065] The remaining charge time prediction model is a mathematical model built using machine learning or data analysis techniques. It aims to predict the remaining charge time of a power battery pack based on input feature data. This model is typically trained using historical data to learn the relationship between various factors (such as battery pack aging and battery temperature) and the remaining charge time.

[0066] Specifically, the remaining charging time prediction model inputs target feature data into a trained algorithm, which then outputs an estimated remaining charging time based on learned patterns and rules. This helps users and systems better manage the charging process, optimize charging strategies, reduce charging wait times, and improve usage efficiency.

[0067] The present invention provides a method for predicting the remaining charge time of a battery, comprising: obtaining target feature data of a target vehicle's power battery pack, wherein the target feature data is data of target features related to the remaining charge time of the power battery pack obtained through feature screening; and inputting the target feature data into a pre-trained remaining charge time prediction model to obtain the remaining charge time of the power battery pack as output by the remaining charge time prediction model. By obtaining target feature data related to the remaining charge time of the battery through feature screening and inputting it into the trained remaining charge time prediction model, the present invention enables the model to more accurately understand and capture key factors affecting charging time, thereby achieving accurate prediction of the remaining charge time of the battery.

[0068] Optional, based on Figure 1 The method shown, such as Figure 2 As shown, a flowchart of an implementation method of feature screening provided by an embodiment of the present invention is provided. The feature screening process may include:

[0069] S200: Obtain a pending feature set of a power battery pack and a corresponding actual remaining charging time, wherein the pending feature set includes a plurality of charging features of the power battery pack.

[0070] The undetermined feature set refers to a set of feature parameters that have not yet been determined when predicting the remaining charging time of the power battery pack, including various charging characteristics that may affect the remaining charging time of the power battery pack.

[0071] Among them, the actual remaining charging time refers to the remaining time actually required to fully charge the power battery pack from the current state under specific charging conditions (that is, under the actual characteristic parameters of each charging characteristic in the pending characteristic set).

[0072] Among them, charging characteristics refer to various technical and environmental parameters that affect the battery charging process.

[0073] In the analysis of power battery packs, in order to obtain the pending feature set and the corresponding actual remaining charging time, multiple key factors need to be considered. The pending feature set is mainly generated by collecting charging characteristics that can characterize the performance and aging status of the battery pack.

[0074] First, during the resting state or charge-discharge cycle, the passivation layer on the surface of the electrode material continuously decomposes and regenerates. At the same time, the electrodes may fracture and undergo lithium plating. These factors can lead to degradation of the anode and cathode, causing the capacity of the lithium battery to age. An aged battery not only reduces its available capacity but also increases its internal resistance. These two factors, combined, can affect charging efficiency and the actual remaining charge time. Therefore, embodiments of the present invention can comprehensively characterize the aging state of the battery pack by collecting state-of-health (SOH) data for the entire battery pack.

[0075] Secondly, battery temperature is a significant factor influencing charging rate. Excessively high temperatures accelerate internal degradation, leading to loss of active materials and increased internal resistance, while excessively low temperatures can cause lithium deposition at the anode. To address the issue of prior art failing to fully account for temperature changes during battery charging, embodiments of the present invention can also collect multiple temperature-related features (e.g., maximum cell temperature, minimum cell temperature, thermal management power, and ambient temperature) to comprehensively consider the effects of the thermal management system's heat exchange, self-generated heat, and external environmental heat exchange on battery temperature, thereby more accurately predicting the remaining charging time.

[0076] Optionally, the charging characteristics include the whole pack SOC, whole pack SOH, battery pack voltage, single cell maximum voltage, single cell minimum voltage, ambient temperature, single cell maximum temperature, single cell minimum temperature and charging loop current characteristics of the power battery pack during the charging process.

[0077] Among them, the whole pack SOC represents the current state of charge of the entire battery pack, usually expressed as a percentage.

[0078] Among them, the whole pack SOH represents the health status of the battery pack, which is usually defined by comparing the current capacity with the capacity of a new battery.

[0079] Among them, the battery pack voltage refers to the output voltage of the entire battery pack, which represents the potential of the battery pack during charging or discharging.

[0080] Among them, the maximum cell voltage refers to the maximum voltage of a single battery cell in the battery pack during the charging process.

[0081] Among them, the lowest cell voltage refers to the lowest voltage of a single battery cell in the battery pack during the charging process.

[0082] The ambient temperature refers to the temperature of the environment in which the battery pack is located.

[0083] The maximum cell temperature refers to the maximum temperature reached by a single battery cell in the battery pack during the charging process.

[0084] The minimum cell temperature refers to the lowest temperature reached by a single battery cell in the battery pack during the charging process.

[0085] Among them, the charging circuit current characteristic refers to the current flowing through the battery pack during the charging process.

[0086] Specifically, the embodiment of the present invention can collect multiple charging characteristics offline, including the whole pack SOC, whole pack SOH, battery pack voltage, single cell maximum voltage, single cell minimum voltage, ambient temperature, single cell maximum temperature, single cell minimum temperature and charging loop current characteristics of the power battery pack during the charging process, to form a pending feature set.

[0087] Furthermore, the embodiment of the present invention can collect offline data based on battery packs produced in the same batch. In this process, the voltage, current and temperature signals will be collected directly, while the whole pack SOC, whole pack SOH and battery pack voltage will be read from the battery management system of the vehicle terminal. At the same time, in order to ensure the accuracy of the remaining charging time and be able to finally display it on the vehicle-side user interface, it is necessary to set the signal acquisition frequency according to the minimum time division unit used by the interface. For example: when the minimum time division unit used by the vehicle-side user interface is 1 second, the signal acquisition frequency is set to be greater than or equal to 1Hz to ensure the real-time and accuracy of the data.

[0088] Optionally, an embodiment of the present invention can perform data cleaning on the feature parameters in the predetermined feature set to identify outliers, and perform online filtering, denoising and data standardization. The specific process can be found in the aforementioned description of data cleaning of the original signal of the target feature of the power battery pack, which will not be repeated here.

[0089] The embodiment of the present invention ensures the validity of the feature parameters in the pending feature set by performing data cleaning on the pending feature set, providing a reliable basis for subsequent feature analysis and model training, thereby improving the prediction accuracy of the trained remaining charging time prediction model.

[0090] S210 , analyzing the correlation between each charging feature in the pending feature set and the actual remaining charging time and the redundancy between each charging feature, and sorting the charging features from large to small according to their comprehensive correlation to obtain a valid feature set.

[0091] This embodiment of the present invention uses statistical analysis methods (such as correlation coefficient calculation and regression analysis) to evaluate the correlation between each charging feature in the undetermined feature set and the actual remaining charging time, thereby obtaining a correlation analysis result for each charging feature. The higher the correlation between a charging feature and the actual remaining charging time, the more important the charging feature is for predicting the remaining charging time.

[0092] In the undetermined feature set, some charging features may have overlapping or similar effects on predicting the remaining charging time. This situation is called redundancy. Embodiments of the present invention can identify redundant features in the undetermined feature set by calculating correlation coefficients between individual charging features or using principal component analysis to obtain redundancy analysis results for each charging feature, thereby reducing unnecessary computational complexity.

[0093] After completing the correlation and redundancy analysis, the embodiment of the present invention will evaluate the comprehensive correlation of the charging features based on the correlation analysis results and redundancy analysis results of each charging feature, and sort these features from high to low according to the comprehensive correlation of each charging feature to form a valid feature set.

[0094] S220 , using the valid feature set to perform model training on the remaining charging time prediction model, and screening out multiple target features related to the remaining charging time of the power battery pack from the valid feature set.

[0095] During the model training process, the remaining charging time prediction model can use the input valid feature set to learn the relationship between charging features and remaining charging time. By continuously adjusting the model parameters, the remaining charging time prediction model will be gradually optimized to reduce the prediction error. At the same time, during the training process, the remaining charging time prediction model will evaluate the impact of different charging feature combinations in the valid feature set on the prediction results. For example: The embodiment of the present invention can test the effects of different charging feature combinations through cross-validation or other model evaluation methods to find the charging feature combination with the smallest prediction error as the multiple target features related to the remaining charging time of the power battery pack.

[0096] The embodiment of the present invention can accurately capture the impact of multiple charging features on the remaining charging time by obtaining the pending feature set of the power battery pack and the corresponding actual remaining charging time. After performing a correlation and redundancy analysis on these charging features, they are sorted according to the comprehensive correlation to form an effective feature set, thereby ensuring that the features used in model training are highly relevant and contain sufficient information. Finally, the remaining charging time prediction model is trained using the effective feature set, and the target features related to the remaining charging time of the battery are screened out, so that the model can more accurately predict the remaining charging time of the battery, improve the accuracy and reliability of the prediction, and provide users with a better charging experience.

[0097] Optional, based on Figure 2 The feature screening process shown in Figure 3 As shown, a flowchart of another implementation of feature screening provided by an embodiment of the present invention is shown, where step S210 may specifically include:

[0098] S300: Calculate the mutual information between each charging feature and the actual remaining charging time.

[0099] To ensure the correlation between the final input signal (i.e., charging characteristics) and the output signal (i.e., remaining charging time) and prevent irrelevant input signals from affecting the accuracy of the prediction model, embodiments of the present invention can use mutual information calculation to screen important input signals. Mutual information is a concept in information theory that measures the uncertainty of one random variable reduced by the knowledge of another random variable, thereby reflecting the amount of information shared between the two random variables (i.e., charging characteristics and remaining charging time).

[0100] Specifically, for two continuous random variables X and Y, their mutual information can be expressed as the joint probability distribution and marginal distribution and The expression is calculated as follows:

[0101]

[0102] For discrete random variables X and Y, the mutual information is defined as:

[0103]

[0104] From these definitions, we can see that the mutual information value represents the degree of association between variables X and Y. When two variables are completely unrelated, their joint probability distribution is zero, and the corresponding mutual information value is also zero. The mutual information value between a random variable and itself is maximum. Therefore, by calculating the mutual information value between the input signal and the remaining charging time, we can assess the correlation between different input signals and the remaining charging time. A larger mutual information value indicates a stronger correlation between the two, while a smaller mutual information value indicates a lower data-level unrelatedness. Through this analysis, we can effectively screen out important input signals related to the remaining charging time.

[0105] S310: Transfer the charging feature with the largest mutual information from the pending feature set to the valid feature set.

[0106] Specifically, the embodiment of the present invention can pre-set a pending feature set S containing all charging features, and then set an effective feature set F for storing selected charging features. The effective feature set F is initially empty. Then, the mutual information between each charging feature in the pending feature set S and the remaining charging time is calculated to find the charging feature with the largest mutual information. , that is, satisfy The maximum charging characteristic. Remove it from the pending feature set S and add it to the valid feature set F.

[0107] S320: Calculate the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, and between each charging feature in the valid feature set.

[0108] It should be noted that the embodiments of the present invention not only focus on the correlation between the input and output signals, but also consider the high correlation (i.e., redundancy) that may exist between the input signals. If highly correlated variables are simultaneously input into the prediction model, it may increase the complexity of the model and the storage and computing burden of the cloud platform. To solve this problem, the embodiments of the present invention calculate the mutual information between the input signals. The specific formula is as follows:

[0109]

[0110] The optimization goal is to select the effective feature set F from the original input sequence X, so that the average mutual information between the input variables in the effective feature set F and the remaining charging time is maximized, and the average mutual information between the input variables and other input variables in the effective feature set F is minimized. That is, the input variables in the effective feature set F are selected and , so that the following conditions are met:

[0111]

[0112] The embodiment of the present invention can sort all charging features by comprehensively considering the correlation and redundancy of input signals, thereby optimizing their application in the prediction model.

[0113] Next, the comprehensive mutual information between each charging feature in the pending feature set S and the actual remaining charging time, as well as each charging feature in the valid feature set F, is calculated. The specific calculation formula is as follows:

[0114]

[0115] in, Represents the number of features in the current valid feature set F. The comprehensive mutual information aims to measure the correlation between each charging feature in the undetermined feature set S and the remaining charging time, while considering its mutual relationship with each charging feature in the valid feature set F.

[0116] S330: Transfer the charging feature with the largest comprehensive mutual information from the pending feature set to the valid feature set.

[0117] Specifically, the embodiment of the present invention can sequentially convert the comprehensive mutual information determined from the undetermined feature set S in each round into Maximum charging feature Remove it from the pending feature set S and add it to the valid feature set F as the first input features.

[0118] S340: Determine whether the undetermined feature set is an empty set. If yes, execute step S320; if no, execute step S350.

[0119] It can be understood that the embodiment of the present invention calculates the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, as well as each charging feature in the valid feature set through multiple rounds of calculation. In each round, at least one charging feature can be selected from the pending feature set and transferred to the valid feature set until all the charging features in the pending feature set are transferred to the valid feature set, and finally the pending feature set is an empty set.

[0120] S350: Obtain a valid feature set sorted from largest to smallest according to comprehensive relevance.

[0121] It can be understood that the charging feature with greater comprehensive correlation is transferred to the effective feature set earlier. Therefore, in the embodiment of the present invention, the charging features in the effective feature set can be sorted according to the order in which the charging features are transferred to the effective feature set, and the effective feature set F is obtained after being sorted from large to small according to comprehensive correlation, that is:

[0122]

[0123] in, is the mth feature of comprehensive correlation in the valid feature set F, n is the total number of features in the valid feature set F, and in this embodiment, n=9.

[0124] The embodiment of the present invention systematically analyzes the correlation between each charging feature in the pending feature set and the actual remaining charging time, as well as the redundancy between the features, and continuously screens and sorts them accordingly, thereby effectively constructing an effective feature set that is rich in information and low in redundancy, thereby improving the prediction performance of the model.

[0125] Optional, in the above Figure 2 On the basis of one or more corresponding embodiments, in another optional embodiment provided by the embodiment of the present invention, step S220 may specifically include:

[0126] The valid feature set is divided into multiple, ascending subsets, each extending the previous one. Each subset is sequentially input into the remaining charge time prediction model to obtain the predicted remaining charge time output by the remaining charge time prediction model. The charging features in the subset with the smallest error between the predicted remaining charge time and the actual remaining charge time are identified as the target features related to the remaining charge time of the power battery pack.

[0127] Specifically, the embodiment of the present invention can divide the effective feature set F into multiple sub-sets with increasing levels, such as " "," "until" " for subsequent analysis. In the model training step, these subsets are sequentially input into the remaining charging time prediction model in the order of comprehensive relevance, and the prediction effect of each subset is calculated.

[0128] Furthermore, the embodiment of the present invention can calculate the root mean square error (MSE) between the prediction result and the actual remaining charging time, using the formula:

[0129]

[0130] in, is the actual remaining charging time at the i-th sampling point, is the predicted remaining charging time for the i-th sampling point. Finally, the subset with the smallest MSE is selected, and the charging features therein are determined as the target features related to the remaining charging time of the power battery pack, thereby ensuring the search for the optimal feature combination and improving the accuracy of the prediction model.

[0131] The embodiment of the present invention determines the target features related to the remaining charging time of the power battery pack through hierarchical feature selection and model evaluation, and can systematically evaluate the contribution of each charging feature in predicting the remaining charging time, thereby accurately screening out the target features that play a key role in predicting the remaining charging time, providing a reliable basis for subsequent model optimization and model prediction.

[0132] Although the operations are depicted in a particular order, this should not be understood as requiring that the operations be performed in the particular order shown or in a sequential order.Multitasking and parallel processing may be advantageous under certain circumstances.

[0133] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0134] In the practical application of the remaining battery charge time prediction method, as the volume of data increases, chips face higher requirements in terms of data storage and computing power. Traditional on-board computing systems are unable to effectively process large amounts of data or deeply mine the information contained in the data. Therefore, the present invention introduces cloud computing technology. The on-board terminal is responsible for collecting and uploading the raw signals, while simultaneously receiving and executing the return signals from the cloud. The cloud is responsible for receiving and processing these raw signals, performing relevant calculations, and then sending a return signal. In this way, large amounts of data can be processed and analyzed more efficiently.

[0135] like Figure 4 As shown, an architectural diagram of a battery remaining charging time prediction system provided by an embodiment of the present invention is provided. The system may include: a cloud, a vehicle terminal, and a battery management system. The cloud and the vehicle terminal are connected via wireless transmission, and the vehicle terminal and the battery management system are connected via a CAN bus.

[0136] The vehicle terminal obtains original feature data of the vehicle's power battery pack from the battery management system, wherein the original feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening.

[0137] The vehicle terminal transmits the original feature data to the cloud.

[0138] After receiving the original feature data transmitted by the vehicle terminal, the cloud cleans the original feature data to obtain the target feature data.

[0139] The cloud inputs the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

[0140] The cloud will feed back the remaining charging time to the vehicle terminal.

[0141] The vehicle terminal transmits the remaining charging time to the battery management system.

[0142] The vehicle terminal obtains raw feature data related to the remaining charge time of the power battery pack from the battery management system. This data is the target feature after feature screening and optimization. The obtained raw feature data is then transmitted to the cloud in real time. After receiving the raw feature data transmitted by the vehicle terminal, the cloud performs data cleaning and preprocessing to extract the target feature data. This target feature data is input into a pre-trained remaining charge time prediction model to calculate the remaining charge time of the power battery pack. After the cloud calculates the remaining charge time, this information is fed back to the vehicle terminal via wireless communication. The vehicle terminal then transmits this remaining charge time to the battery management system for appropriate management and decision-making.

[0143] Corresponding to the above method embodiment, the embodiment of the present invention also provides a battery remaining charging time prediction device, the structure of which is as follows: Figure 5 As shown, it may include: a target characteristic data obtaining unit 10 and a remaining charging time obtaining unit 20.

[0144] The target feature data obtaining unit 10 is used to obtain target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening.

[0145] The remaining charging time obtaining unit 20 is used to input the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

[0146] Optionally, the battery remaining charging time prediction device may further include: a feature screening unit, wherein the feature screening unit includes: a power package data acquisition subunit, a valid feature set acquisition subunit and a target feature screening subunit.

[0147] The power pack data acquisition subunit is used to obtain a pending feature set of the power battery pack and the corresponding actual remaining charging time, wherein the pending feature set includes multiple charging features of the power battery pack.

[0148] The effective feature set acquisition subunit is used to analyze the correlation between each charging feature in the pending feature set and the actual remaining charging time and the redundancy between each charging feature, and sort each charging feature from large to small according to the comprehensive correlation to obtain the effective feature set.

[0149] The target feature screening subunit is used to use the valid feature set to train the remaining charging time prediction model, and screen out multiple target features related to the remaining charging time of the power battery pack in the valid feature set.

[0150] Optionally, the effective feature set acquisition subunit is specifically used to calculate the mutual information between each charging feature and the actual remaining charging time; transfer the charging feature with the largest mutual information from the pending feature set to the effective feature set; calculate the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, as well as with each charging feature in the effective feature set; transfer the charging feature with the largest comprehensive mutual information from the pending feature set to the effective feature set; determine whether the pending feature set is an empty set. If so, execute the step of calculating the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, as well as with each charging feature in the effective feature set. If not, obtain the effective feature set sorted from large to small according to the comprehensive correlation.

[0151] Optionally, a target feature screening subunit is specifically used to divide the valid feature set into multiple hierarchically increasing subsets, each subset being an extension of the previous subset; inputting each subset into the remaining charging time prediction model in turn to obtain the predicted remaining charging time output by the remaining charging time prediction model; and confirming the charging feature in the subset with the smallest error between the predicted remaining charging time and the actual remaining charging time as the target feature related to the remaining charging time of the power battery pack.

[0152] Optionally, the charging characteristics include the whole pack SOC, whole pack SOH, battery pack voltage, single cell maximum voltage, single cell minimum voltage, ambient temperature, single cell maximum temperature, single cell minimum temperature and charging loop current characteristics of the power battery pack during the charging process.

[0153] Optionally, the target characteristic data is data obtained by cleaning the original signal of the target characteristic of the power battery pack.

[0154] The present invention provides a device for predicting the remaining charge time of a battery. The device is used to: obtain target feature data of a target vehicle's power battery pack, wherein the target feature data is data of target features related to the remaining charge time of the power battery pack obtained through feature screening; and input the target feature data into a pre-trained remaining charge time prediction model to obtain the remaining charge time of the power battery pack as output by the remaining charge time prediction model. By obtaining target feature data related to the remaining charge time of the battery through feature screening and inputting it into the trained remaining charge time prediction model, the present invention enables the model to more accurately understand and capture key factors affecting charging time, thereby achieving accurate prediction of the remaining charge time of the battery.

[0155] Regarding the apparatus in the above embodiment, the specific manner in which each unit performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0156] The battery remaining charging time prediction device includes a processor and a memory. The above-mentioned target feature data acquisition unit 10 and the remaining charging time acquisition unit 20 are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions.

[0157] The processor includes a kernel, which retrieves the corresponding program unit from memory. One or more kernels can be configured. By adjusting kernel parameters, target feature data related to the remaining battery charge time is obtained through feature screening and input into a trained remaining charge time prediction model. This enables the model to more accurately understand and capture the key factors affecting charging time, thereby achieving a precise prediction of the remaining battery charge time.

[0158] An embodiment of the present invention provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the method for predicting the remaining charging time of a battery is implemented.

[0159] An embodiment of the present invention provides a processor, which is used to run a program, wherein the method for predicting the remaining battery charging time is executed when the program is run.

[0160] like Figure 6 As shown, an embodiment of the present invention provides an electronic device 1000, which includes at least one processor 1001, at least one memory 1002 connected to the processor 1001, and a bus 1003. The processor 1001 and the memory 1002 communicate with each other via the bus 1003. The processor 1001 is configured to call program instructions in the memory 1002 to execute the above-described method for predicting the remaining battery charge time. The electronic device herein may be a server, a PC, a PAD, a mobile phone, etc.

[0161] The present invention also provides a computer program product, which, when executed on an electronic device, is suitable for executing a program initialized with the steps of the method for predicting the remaining battery charging time.

[0162] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatuses, electronic devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable device to produce a machine, so that the instructions executed by the processor of the computer or other programmable device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0163] In a typical configuration, an electronic device includes one or more processors (CPUs), a memory, and a bus. The electronic device may also include an input / output interface, a network interface, and the like.

[0164] Memory may include non-permanent memory in a computer-readable medium, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip. Memory is an example of a computer-readable medium.

[0165] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0166] In the description of the present invention, it should be understood that if the terms "up", "down", "front", "back", "left" and "right" are used to indicate directions or positional relationships, they are based on the directions or positional relationships shown in the accompanying drawings. They are only used to facilitate the description of the present invention and simplify the description, and do not indicate or imply that the positions or elements referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limitations of the present invention.

[0167] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. It should also be noted that the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, commodity, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, commodity, or device comprising the element.

[0168] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0169] The above are merely embodiments of the present invention and are not intended to limit the present invention. It will be apparent to those skilled in the art that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are intended to be within the scope of the present invention.

Claims

1. A method for predicting the remaining charging time of a battery, characterized in that: include: Obtaining target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to a remaining charging time of the power battery pack obtained through feature screening; The target feature data is input into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

2. The method according to claim 1, characterized in that The feature screening process includes: Obtaining a pending feature set of the power battery pack and a corresponding actual remaining charging time, wherein the pending feature set includes a plurality of charging features of the power battery pack; Analyzing the correlation between each charging feature and the actual remaining charging time and the redundancy between each charging feature in the pending feature set, and sorting the charging features from largest to smallest according to their comprehensive correlation to obtain a valid feature set; The remaining charging time prediction model is trained using the effective feature set, and a plurality of target features related to the remaining charging time of the power battery pack are screened out from the effective feature set.

3. The method according to claim 2, characterized in that The analysis of the correlation between each charging feature and the actual remaining charging time and the redundancy between each charging feature in the undetermined feature set is performed, and the charging features are sorted from largest to smallest according to their comprehensive correlation to obtain a valid feature set, including: respectively calculating the mutual information between each of the charging characteristics and the actual remaining charging time; Transferring the charging feature with the largest mutual information from the pending feature set to the valid feature set; Calculating comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, and between each charging feature in the valid feature set; Transferring the charging feature with the largest comprehensive mutual information from the pending feature set to the valid feature set; Determine whether the pending feature set is an empty set. If so, perform the step of calculating the comprehensive mutual information between each charging feature in the charging feature set and the actual remaining charging time, and each charging feature in the valid feature set. If not, obtain the valid feature set sorted from large to small according to the comprehensive correlation.

4. The method according to claim 2, characterized in that The method of using the effective feature set to train the remaining charging time prediction model and screening out multiple target features related to the remaining charging time of the power battery pack from the effective feature set includes: Dividing the effective feature set into a plurality of hierarchically increasing subsets, each subset being an extension of the previous subset; Inputting each subset into the remaining charging time prediction model in turn to obtain the predicted remaining charging time output by the remaining charging time prediction model; The charging feature in the subset with the smallest error between the predicted remaining charging time and the actual remaining charging time is confirmed as a target feature related to the remaining charging time of the power battery pack.

5. The method according to any one of claims 2 to 4, characterized in that The charging characteristics include the whole pack SOC, whole pack SOH, battery pack voltage, single cell maximum voltage, single cell minimum voltage, ambient temperature, single cell maximum temperature, single cell minimum temperature and charging circuit current characteristics of the power battery pack during the charging process.

6. The method according to claim 1, characterized in that The target feature data is data obtained by cleaning the original signal of the target feature of the power battery pack.

7. A battery remaining charging time prediction device, characterized in that: include: Target characteristic data obtaining unit and remaining charging time obtaining unit, The target feature data obtaining unit is configured to obtain target feature data of a power battery pack of a target vehicle, wherein the target feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening; The remaining charging time obtaining unit is used to input the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model.

8. A battery remaining charging time prediction system, characterized in that: include: A cloud, a vehicle terminal, and a battery management system, wherein the cloud and the vehicle terminal are connected via wireless transmission, and the vehicle terminal and the battery management system are connected via a CAN bus; The vehicle terminal obtains original feature data of the vehicle's power battery pack from the battery management system, wherein the original feature data is target feature data related to the remaining charging time of the power battery pack obtained through feature screening; The vehicle terminal transmits the original feature data to the cloud; After receiving the original feature data transmitted by the vehicle terminal, the cloud performs data cleaning on the original feature data to obtain target feature data; The cloud inputs the target feature data into a pre-trained remaining charging time prediction model to obtain the remaining charging time of the power battery pack output by the remaining charging time prediction model; The cloud feeds back the remaining charging time to the vehicle terminal; The vehicle terminal transmits the remaining charging time to the battery management system.

9. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the method for predicting the remaining battery charging time according to any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: The electronic device includes at least one processor, and at least one memory and bus connected to the processor; wherein the processor and the memory communicate with each other via the bus; the processor is used to call program instructions in the memory to execute the battery remaining charging time prediction method according to any one of claims 1 to 6.

Images