A battery remaining charging time prediction method, device and electronic equipment

Abstract

The application provides a battery remaining charging time prediction method and device and electronic equipment. The method comprises the following steps: obtaining charging environment information and battery attribute information of a battery to be measured; determining prediction characteristics of the battery to be measured according to the charging environment information and the battery attribute information; and determining a remaining charging time prediction result of the battery to be measured based on a preset fusion model and the prediction characteristics of the battery to be measured. The method provided by the above scheme can improve the accuracy of the prediction result by predicting the remaining charging time of the battery to be measured based on the fusion model and in combination with the charging environment information and the battery attribute information of the battery to be measured.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus and electronic device for predicting the remaining charging time of a battery. Background Technology

[0002] Currently, battery charging information is one of the most important pieces of information for electric vehicle users. Predicting the remaining charging time allows customers to know how long it will take for the battery to fully charge, so they can plan their travel time accordingly.

[0003] In existing technologies, the charging time required to fully charge the battery from its current SOC is typically calculated based on the average charging time of each charge, which is characterized by the battery's SOC during the vehicle's historical charging records.

[0004] However, since the charging environment and battery usage vary each time a vehicle is charged, the accuracy of the average SOC charging time obtained by existing technologies cannot be guaranteed, and therefore the accuracy of the final prediction results cannot be guaranteed. Summary of the Invention

[0005] This application provides a method, apparatus, and electronic device for predicting the remaining charging time of a battery, in order to overcome the shortcomings of existing technologies, such as the inability to guarantee the accuracy of prediction results.

[0006] The first aspect of this application provides a method for predicting the remaining charging time of a battery, including:

[0007] Obtain charging environment information and battery attribute information of the battery under test;

[0008] Based on the charging environment information and battery attribute information, the predicted characteristics of the battery under test are determined;

[0009] Based on a preset fusion model, the remaining charging time prediction result of the battery under test is determined according to the prediction characteristics of the battery under test.

[0010] Optionally, the step of determining the remaining charging time prediction result of the battery under test based on the prediction characteristics of the battery under test using a preset fusion model includes:

[0011] Based on the preset first LightGBM model, the first intermediate prediction result is determined according to the prediction characteristics of the battery under test;

[0012] The first intermediate prediction result is used as the first new feature and added to the prediction feature to obtain the first fused prediction feature;

[0013] Based on the preset first XGBoost model, and according to the first fusion prediction features, the second intermediate prediction result is determined;

[0014] The second intermediate prediction result is used as a second new feature and added to the prediction feature to obtain the second fused prediction feature;

[0015] Based on the preset second LightGBM model, the prediction result of the LightGBM model is determined according to the second fusion prediction feature;

[0016] Based on the preset second XGBoost model, the XGBoost model prediction result is determined according to the first fusion prediction feature;

[0017] Based on the preset neural network model, the prediction result of the neural network model is determined according to the prediction features;

[0018] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0019] Optionally, determining the predicted remaining charging time of the battery under test based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model includes:

[0020] Based on the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined respectively.

[0021] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, as well as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0022] Optionally, determining the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model based on their prediction accuracies includes:

[0023] Based on several preset test sets, the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model is verified.

[0024] For any of the aforementioned models, the error rate of the model in each of the aforementioned test sets is determined based on the prediction accuracy verification results corresponding to each of the aforementioned test sets.

[0025] With the goal of reducing the error rate difference between the various test sets, the error rate of the model in each test set is corrected to obtain the error rate feature set of the model.

[0026] Based on the error rate feature set of each model, several model weight feature sets are determined; wherein, the model weight feature set includes the error rate features generated by each model on the same test set;

[0027] Based on the aforementioned set of model weight features, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined.

[0028] Optionally, determining the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model based on the plurality of model weight feature sets includes:

[0029] Based on the model weight feature set, multiple sets of candidate prediction weights are determined;

[0030] Based on the aforementioned test sets, the prediction accuracy of each group of candidate prediction weights when applied to the second LightGBM model, the second XGBoost model, and the neural network model is verified.

[0031] The candidate prediction weights with the highest prediction accuracy are determined as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

[0032] Optional, also includes:

[0033] Acquire battery charging data from multiple vehicles; wherein, the battery charging data includes charging environment information, battery attribute information, and timestamps corresponding to each information indicator during the battery charging process;

[0034] The battery charging data is preprocessed to obtain the model training set;

[0035] Using the model training set, a fusion model is trained to obtain the preset fusion model.

[0036] Optionally, the predicted characteristics of the battery under test include at least the vehicle charging method, charging current, charging voltage, ambient temperature, state of charge, and number of battery charge-discharge cycles.

[0037] A second aspect of this application provides a battery remaining charging time prediction device, comprising:

[0038] The acquisition module is used to acquire charging environment information and battery attribute information of the battery under test;

[0039] The determination module is used to determine the predicted characteristics of the battery under test based on the charging environment information and battery attribute information.

[0040] The prediction module is used to determine the predicted remaining charging time of the battery under test based on a preset fusion model and the predicted characteristics of the battery under test.

[0041] Optionally, the prediction module is specifically used for:

[0042] Based on the preset first LightGBM model, the first intermediate prediction result is determined according to the prediction characteristics of the battery under test;

[0043] The first intermediate prediction result is used as the first new feature and added to the prediction feature to obtain the first fused prediction feature;

[0044] Based on the preset first XGBoost model, and according to the first fusion prediction features, the second intermediate prediction result is determined;

[0045] The second intermediate prediction result is used as a second new feature and added to the prediction feature to obtain the second fused prediction feature;

[0046] Based on the preset second LightGBM model, the prediction result of the LightGBM model is determined according to the second fusion prediction feature;

[0047] Based on the preset second XGBoost model, the XGBoost model prediction result is determined according to the first fusion prediction feature;

[0048] Based on the preset neural network model, the prediction result of the neural network model is determined according to the prediction features;

[0049] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0050] Optionally, the prediction module is specifically used for:

[0051] Based on the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined respectively.

[0052] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, as well as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0053] Optionally, the prediction module is specifically used for:

[0054] Based on several preset test sets, the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model is verified.

[0055] For any of the aforementioned models, the error rate of the model in each of the aforementioned test sets is determined based on the prediction accuracy verification results corresponding to each of the aforementioned test sets.

[0056] With the goal of reducing the error rate difference between the various test sets, the error rate of the model in each test set is corrected to obtain the error rate feature set of the model.

[0057] Based on the error rate feature set of each model, several model weight feature sets are determined; wherein, the model weight feature set includes the error rate features generated by each model on the same test set;

[0058] Based on the aforementioned set of model weight features, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined.

[0059] Optionally, the prediction module is specifically used for:

[0060] Based on the model weight feature set, multiple sets of candidate prediction weights are determined;

[0061] Based on the aforementioned test sets, the prediction accuracy of each group of candidate prediction weights when applied to the second LightGBM model, the second XGBoost model, and the neural network model is verified.

[0062] The candidate prediction weights with the highest prediction accuracy are determined as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

[0063] Optionally, the device further includes:

[0064] The model training module is used to acquire battery charging data from multiple vehicles; wherein, the battery charging data includes charging environment information, battery attribute information, and timestamps corresponding to each information indicator during the battery charging process; the battery charging data is preprocessed to obtain a model training set; the fusion model is trained using the model training set to obtain the preset fusion model.

[0065] Optionally, the predicted characteristics of the battery under test include at least the vehicle charging method, charging current, charging voltage, ambient temperature, state of charge, and number of battery charge-discharge cycles.

[0066] A third aspect of this application provides an electronic device, comprising: at least one processor and a memory;

[0067] The memory stores computer-executed instructions;

[0068] The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the method described in the first aspect above and various possible designs of the first aspect.

[0069] The fourth aspect of this application provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the method described in the first aspect above and various possible designs of the first aspect.

[0070] The technical solution of this application has the following advantages:

[0071] This application provides a method, apparatus, and electronic device for predicting the remaining charging time of a battery. The method includes: acquiring charging environment information and battery attribute information of the battery under test; determining prediction characteristics of the battery under test based on the charging environment information and battery attribute information; and determining the prediction result of the remaining charging time of the battery under test based on a preset fusion model and the prediction characteristics of the battery under test. The method provided above improves the accuracy of the prediction result by predicting the remaining charging time of the battery under test by combining the charging environment information and battery attribute information of the battery under test with a fusion model. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0073] Figure 1 This is a schematic diagram of the battery remaining charging time prediction system based on the embodiments of this application;

[0074] Figure 2 A flowchart illustrating the battery remaining charging time prediction method provided in this application embodiment;

[0075] Figure 3 A flowchart illustrating an exemplary battery remaining charging time prediction method provided in an embodiment of this application;

[0076] Figure 4 A schematic diagram of the battery remaining charging time prediction device provided in an embodiment of this application;

[0077] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0078] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the present disclosure in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0080] Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. In the following descriptions of embodiments, "a plurality of" means two or more, unless otherwise explicitly defined.

[0081] In existing technologies, the charging time required to fully charge the battery from its current SOC is typically calculated based on the average charging time of each charge, which represents the battery's SOC during historical charging records. Alternatively, charging time can be predicted based on physical characteristics, using rated power, charging voltage, and the SOC value. Machine learning models, such as random forests, are also used to predict battery charging time. However, because the charging environment and battery usage vary with each vehicle charge, the accuracy of the average SOC charging time obtained by existing technologies cannot be guaranteed, thus compromising the accuracy of the final prediction. Even with the introduction of machine learning models, different models have different training principles and learn different knowledge from historical data. The varying learning abilities of different models under different conditions and in different aspects result in weak overall model performance.

[0082] To address the aforementioned issues, this application provides a method, apparatus, and electronic device for predicting remaining battery charging time. The method includes: acquiring charging environment information and battery attribute information of the battery under test; determining prediction characteristics of the battery under test based on the charging environment information and battery attribute information; and determining the predicted remaining charging time of the battery under test based on a preset fusion model and the predicted characteristics of the battery under test. The method provided above improves the accuracy of the prediction results by predicting the remaining charging time of the battery under test based on a fusion model and combining the charging environment information and battery attribute information of the battery under test.

[0083] The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0084] First, the structure of the battery remaining charging time prediction system on which this application is based will be described:

[0085] The battery remaining charging time prediction method, apparatus, and electronic equipment provided in this application are applicable to predicting the remaining charging time of electric vehicle batteries. Figure 1 The diagram shows the structure of the battery remaining charging time prediction system based on the embodiments of this application. It mainly includes a battery under test, a data acquisition device, and a battery remaining charging time prediction device. Specifically, the data acquisition device collects charging environment information and battery attribute information of the battery under test and sends the obtained information to the battery remaining charging time prediction device, which predicts the remaining charging time of the battery under test based on the obtained information.

[0086] This application provides a method for predicting the remaining charging time of a battery, used to predict the remaining charging time of an electric vehicle battery. The execution subject of this application embodiment is an electronic device, such as a server, desktop computer, laptop computer, tablet computer, and other electronic devices that can be used to predict the remaining charging time of a battery.

[0087] like Figure 2 The diagram shown is a flowchart illustrating a battery remaining charging time prediction method provided in an embodiment of this application. The method includes:

[0088] Step 201: Obtain the charging environment information and battery attribute information of the battery under test.

[0089] Among them, the charging environment information can characterize the impact of factors such as power supply conditions and weather on battery charging, while the battery attribute information can characterize the state of charge and usage of the battery itself.

[0090] Step 202: Determine the predicted characteristics of the battery under test based on the charging environment information and battery attribute information.

[0091] Specifically, data analysis can be performed on charging environment information and battery attribute information to obtain various predictive characteristics of the battery under test. These predictive characteristics include at least the vehicle charging method, charging current, charging voltage, ambient temperature, state of charge, and number of battery charge-discharge cycles.

[0092] Step 203: Based on the preset fusion model, determine the predicted remaining charging time of the battery under test according to the predicted characteristics of the battery under test.

[0093] It should be noted that the preset fusion model can be built based on a variety of different machine learning algorithms, so as to use different machine learning algorithms to jointly predict the remaining charging time of the battery under test, thereby further improving the accuracy of the prediction results.

[0094] Based on the above embodiments, in order to improve the accuracy of the prediction results, as an implementable approach, in one embodiment, based on a preset fusion model, the remaining charging time prediction result of the battery under test is determined according to the prediction characteristics of the battery under test, including:

[0095] Step 2031: Based on the preset first LightGBM model, determine the first intermediate prediction result according to the prediction characteristics of the battery under test;

[0096] Step 2032: The first intermediate prediction result is used as the first new feature and added to the prediction features to obtain the first fused prediction feature;

[0097] Step 2033: Based on the preset first XGBoost model, determine the second intermediate prediction result according to the first fusion prediction features;

[0098] Step 2034: Add the second intermediate prediction result as the second new feature to the prediction features to obtain the second fused prediction feature;

[0099] Step 2035: Based on the preset second LightGBM model, determine the prediction result of the LightGBM model according to the second fusion prediction features;

[0100] Step 2036: Based on the preset second XGBoost model, determine the XGBoost model prediction result according to the first fusion prediction feature;

[0101] Step 2037: Based on the preset neural network model, determine the prediction result of the neural network model according to the prediction features;

[0102] Step 2038: Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, determine the predicted result of the remaining charging time of the battery under test.

[0103] It should be noted that although both the first XGBoost model and the second XGBoost model use the first fused prediction feature as input features to predict the remaining charging time of the battery under test, the hyperparameters of the two models are different. Therefore, there are some differences between the second intermediate prediction results and the XGBoost model prediction results.

[0104] Specifically, in one embodiment, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model can be determined based on their prediction accuracies. The remaining charging time prediction result of the battery under test can be determined based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, as well as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

[0105] For example, such as Figure 3The diagram illustrates an exemplary battery remaining charging time prediction method provided in this application. First, prediction features (feature1), including state of charge, battery charge / discharge cycle count, charging current, charging voltage, and ambient temperature, are input into a preset first LightGBM model to obtain a corresponding first intermediate prediction result (LGB1). Then, LGB1 is added as a first new feature (new feature 1) to feature1, resulting in a first fused prediction feature (feature2). Next, based on a first XGBoost model and a second XGBoost model, prediction is performed according to feature2, yielding a second new feature (new feature 2) and an XGBoost model prediction result (XGB2). New feature 2 is then added to feature1 to obtain a second fused prediction feature (feature3). Finally, based on a second LightGBM model and feature2, the LightGBM model prediction result is determined, and based on feature1, the neural network model prediction result (LSTM) is determined. Finally, based on the prediction weights of the second LightGBM model, the second XGBoost model, and the neural network model, the prediction results of the LightGBM model, the XGBoost model, and the neural network model are weighted and fused to obtain the prediction result of the remaining charging time of the battery under test.

[0106] Specifically, in one embodiment, since the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model directly affect the accuracy of the remaining charging time prediction result, in order to further improve the accuracy of the prediction result, the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model can be verified based on several preset test sets; for any model, the error rate of the model in each test set is determined according to the prediction accuracy verification results corresponding to each test set; with the goal of reducing the error rate difference between each test set, the error rate of the model in each test set is corrected to obtain the error rate feature set of the model; several model weight feature sets are determined according to the error rate feature sets of each model; wherein, the model weight feature sets include the error rate features generated by each model for the same test set; and the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined according to the several model weight feature sets.

[0107] Specifically, the second LightGBM model, the second XGBoost model, and the neural network model can be applied to several test sets to verify prediction accuracy. The error rate of each model on any test set can be calculated using the following formula:

[0108]

[0109] Where MAPE represents the error rate, y i Represents the true value. This represents the predicted value. To reduce the impact of models with large MAPE on the final prediction results, the goal can be to reduce the error rate difference between different test sets, thus correcting the error rate of the model in each test set. Taking three test sets as an example, the error rates of the model in each test set can be sorted from smallest to largest, and then 0.4, 0.5, and 0.6 can be subtracted successively to obtain the three error rate features of the model, i.e., the error rate feature set.

[0110] Furthermore, model weight feature sets ΔMAPE1{mape1,mape2,mape3}, ΔMAPE2{mape1,mape2,mape3}, and ΔMAPE3{mape1,mape2,mape3} can be constructed for each test set, where mape1 represents the MAPE value calculated by the second XGBoost model, mape2 represents the MAPE value of the neural network model, and mape3 represents the MAPE value of the second LightGBM model.

[0111] Furthermore, in one embodiment, multiple sets of candidate prediction weights can be determined based on the model weight feature set; the prediction accuracy of each set of candidate prediction weights when applied to the second LightGBM model, the second XGBoost model, and the neural network model can be verified based on several test sets; and the set of candidate prediction weights with the highest prediction accuracy can be determined as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

[0112] Specifically, the candidate prediction weights corresponding to the weight feature sets of each model can be calculated using the following formula:

[0113]

[0114] The final prediction result is a1*GB2+a2*STM+a3*GB2.

[0115] Specifically, in order to ensure the accuracy of the final prediction results of the fusion model, the candidate prediction weights corresponding to each model weight feature set can be applied to the second LightGBM model, the second XGBoost model, and the neural network model, and the prediction accuracy of their prediction results can be verified. Finally, the set of candidate prediction weights with the highest prediction accuracy is taken as the target prediction weights, that is, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined, such as a1 = 0.23, a2 = 0.35, and a3 = 0.42.

[0116] Based on the above embodiments, as an implementable approach, in one embodiment, the method further includes:

[0117] Step 301: Obtain battery charging data from multiple vehicles; wherein, the battery charging data includes charging environment information, battery attribute information, and timestamps corresponding to each information indicator during the battery charging process.

[0118] Step 302: Perform data preprocessing on the battery charging data to obtain the model training set;

[0119] Step 303: Use the model training set to train the fusion model and obtain the preset fusion model.

[0120] It should be noted that after preprocessing the battery charging data, the data samples can be divided into a model training set and a test set. For example, the sample data outside the model training set can be divided into three test sets.

[0121] Specifically, battery charging data from dozens of electric vehicles over the past two years can be collected from the BMS system. This data includes information about the battery itself, such as the vehicle charging method, charging current, charging voltage, SOC, and number of charge / discharge cycles, as well as external information such as ambient temperature.

[0122] Furthermore, to make the obtained fusion model more accurate, the battery charging data is preprocessed as follows:

[0123] 1) Data deduplication: Remove duplicate sample data with completely identical features from the original data and keep only one sample data.

[0124] 2) Arrange the raw data in chronological order.

[0125] 3) Abnormal data handling: Based on the upper / lower quartile method of the box plot, delete the entire sample data containing data that exceeds the normal range.

[0126] 4) Handling missing data: When there are 5 or fewer consecutive missing data entries in a field, linear imputation is used to fill in the missing data. When there are many consecutive missing data entries in a field (more than 5 consecutive entries), the sample data containing these missing values ​​are directly deleted.

[0127] 5) Perform one-hot encoding on category features. For example, the feature of "charging method" has two types: "fast charging" and "slow charging". After one-hot encoding, "fast charging" and "slow charging" are represented by two vectors [1,0] and [0,1], respectively.

[0128] 6) Extract the charging segment data from the data cleaned by the above steps, that is, only include the battery charging process data.

[0129] 7) Data labeling: Calculate the time (in minutes) from the SOC value of each sample to the next SOC of 100%, which is the actual remaining charging time. Use this field data as the target variable of the model.

[0130] Specifically, the preset fusion model includes a first LightGBM model, a second LightGBM model, a first XGBoost model, a second XGBoost model, and a neural network model. XGBoost and LightGBM are decision tree models, and the neural network model uses the LSTM algorithm. These two types of models have significantly different principles and their prediction results have low correlation. Fusion of these models is beneficial to improving the accuracy of the prediction results.

[0131] The model training set can be used to construct an n*6 matrix (where n represents the number of samples and 6 represents the number of features) using vehicle charging method, charging current, charging voltage, SOC, battery charge / discharge cycle count, and ambient temperature as input features. This n*6 matrix can be input into the initial state of the first LightGBM model for training, resulting in a trained first LightGBM model. The final hyperparameters of the first LightGBM model are selected as follows:

[0132] {"objective":"regression","n_estimators":5500,"min_child_samples":20,"num_leaves":20,"learning_rate":0.05,"feature_fraction":0.8,"subsample":0.5}.

[0133] Furthermore, the prediction results of the first LightGBM model are used as new features to construct training set 2, resulting in an n*8 matrix. This matrix is ​​then input into the initial state of the first and second XGBoost models for training, yielding the trained first and second XGBoost models. Finally, the hyperparameters of the first XGBoost model are selected as follows:

[0134] {"objective":"reg:linear","n_estimators":8000,"min_child_weight":1,"num_leaves":20,"learning_rate":0.01,"max_depth":6,"subsample":0.6,"colsample_bytree":0.4,"colsample_bylevel":1}.

[0135] The hyperparameters of the second XGBoost model are selected as follows:

[0136] {"objective":"reg:linear","n_estimators":8000,"min_child_weight":1,"num_leaves":26,"learning_rate":0.01,"max_depth":5,"subsample":0.4,"colsample_bytree":0.4,"colsample_bylevel":1}.

[0137] Furthermore, the prediction results of the first XGBoost model are applied to the training set, i.e., the output results are used as new features to construct an n*8 input matrix, which is then input into the second LightGBM model in its initial state for training. This results in a trained second LightGBM model. Finally, the hyperparameters of the second LightGBM model are selected as follows:

[0138] {"objective":"regression","n_estimators":6000,"min_child_samples":20,"num_leaves":20,"learning_rate":0.01,"feature_fraction":0.6,"subsample":0.4}.

[0139] Furthermore, the initially obtained n*7 matrix is ​​used as the training set to train the LSTM model (neural network model). To make the model converge faster, the data is normalized before being input into the LSTM model. x im This represents the sequence after the feature sequence has been normalized. Represents the original feature sequence. This represents the minimum value of the feature sequence in the sample. This represents the maximum value of the feature sequence in the sample. The normalized feature matrix is ​​input into the constructed LSTM model. The LSTM network structure consists of: an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to input samples; the LSTM layer has 200 neurons, and the activation function is... The activation function of the fully connected layer is set to Set mse as the loss function and adam as the model optimizer to train an LSTM model.

[0140] The battery remaining charging time prediction method provided in this application obtains the charging environment information and battery attribute information of the battery under test; determines the prediction characteristics of the battery under test based on the charging environment information and battery attribute information; and determines the prediction result of the remaining charging time of the battery under test based on a preset fusion model and the prediction characteristics of the battery under test. The method provided above improves the accuracy of the prediction results by predicting the remaining charging time of the battery under test based on a fusion model and combining the charging environment information and battery attribute information of the battery under test. Furthermore, by using different types of machine learning algorithms to train different models and comprehensively considering the advantages of different models, multiple models are organically integrated to ensure the accuracy of vehicle remaining charging time prediction, which is beneficial to improving the user's driving experience.

[0141] This application provides a battery remaining charging time prediction device for executing the battery remaining charging time prediction method provided in the above embodiments.

[0142] like Figure 4 The diagram shown is a structural schematic of a battery remaining charging time prediction device provided in an embodiment of this application. The battery remaining charging time prediction device 40 includes: an acquisition module 401, a determination module 402, and a prediction module 403.

[0143] The module includes an acquisition module for acquiring charging environment information and battery attribute information of the battery under test; a determination module for determining the predictive characteristics of the battery under test based on the charging environment information and battery attribute information; and a prediction module for determining the predicted remaining charging time of the battery under test based on a preset fusion model and the predicted characteristics of the battery under test.

[0144] Specifically, in one embodiment, the prediction module is specifically used for:

[0145] Based on the preset first LightGBM model, the first intermediate prediction result is determined according to the prediction characteristics of the battery under test;

[0146] The first intermediate prediction result is used as the first new feature and added to the prediction features to obtain the first fused prediction feature;

[0147] Based on the preset first XGBoost model, the second intermediate prediction result is determined according to the first fusion prediction feature;

[0148] The second intermediate prediction result is used as the second new feature and added to the prediction features to obtain the second fused prediction feature;

[0149] Based on the preset second LightGBM model, the prediction result of the LightGBM model is determined according to the second fusion prediction feature;

[0150] Based on the preset second XGBoost model, the prediction result of the XGBoost model is determined according to the first fusion prediction feature;

[0151] Based on the preset neural network model, the prediction result of the neural network model is determined according to the prediction features;

[0152] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0153] Specifically, in one embodiment, the prediction module is specifically used for:

[0154] Based on the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined respectively.

[0155] Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, as well as the prediction weights of the second LightGBM model, the second XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

[0156] Specifically, in one embodiment, the prediction module is specifically used for:

[0157] Based on several pre-set test sets, the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model is verified.

[0158] For any given model, the error rate of the model in each test set is determined based on the prediction accuracy verification results corresponding to each test set.

[0159] With the goal of reducing the error rate difference between different test sets, the error rate of the model in each test set is corrected to obtain the error rate feature set of the model.

[0160] Based on the error rate feature set of each model, several model weight feature sets are determined; wherein, the model weight feature set includes the error rate features generated by each model on the same test set;

[0161] Based on several sets of model weight features, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined.

[0162] Specifically, in one embodiment, the prediction module is specifically used for:

[0163] Based on the model weight feature set, determine multiple sets of candidate prediction weights;

[0164] Based on several test sets, the prediction accuracy of each group of candidate prediction weights when applied to the second LightGBM model, the second XGBoost model, and the neural network model was verified.

[0165] The candidate prediction weights with the highest prediction accuracy are determined as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

[0166] Specifically, in one embodiment, the device further includes:

[0167] The model training module is used to acquire battery charging data from multiple vehicles. The battery charging data includes charging environment information, battery attribute information, and timestamps corresponding to each information indicator during the battery charging process. The battery charging data is preprocessed to obtain a model training set. The model training set is used to train a fusion model to obtain a preset fusion model.

[0168] Specifically, in one embodiment, the predicted characteristics of the battery under test include at least the vehicle charging method, charging current, charging voltage, ambient temperature, state of charge, and number of battery charge-discharge cycles.

[0169] Regarding the battery remaining charging time prediction device in this embodiment, the specific way in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated here.

[0170] The battery remaining charging time prediction device provided in this application embodiment is used to execute the battery remaining charging time prediction method provided in the above embodiment. Its implementation method and principle are the same, and will not be described again.

[0171] This application provides an electronic device for executing the battery remaining charging time prediction method provided in the above embodiments.

[0172] like Figure 5 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application. The electronic device 50 includes at least one processor 51 and a memory 52.

[0173] The memory stores computer-executable instructions; at least one processor executes the computer-executable instructions stored in the memory, causing the at least one processor to execute the battery remaining charging time prediction method provided in the above embodiment.

[0174] This application provides an electronic device for executing the battery remaining charging time prediction method provided in the above embodiments. Its implementation method and principle are the same, and will not be described again.

[0175] This application provides a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the battery remaining charging time prediction method provided in any of the above embodiments.

[0176] The storage medium containing computer-executable instructions in the embodiments of this application can be used to store computer-executable instructions for the battery remaining charging time prediction method provided in the foregoing embodiments. Its implementation method and principle are the same, and will not be described again.

[0177] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0178] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0179] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0180] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0181] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is merely an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0182] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for predicting the remaining charging time of a battery, characterized in that, include: Acquire charging environment information and battery attribute information of the battery under test; wherein, the charging environment information represents the power supply conditions of the battery under test and the impact of weather on battery charging, and the battery attribute information represents the state of charge and battery usage of the battery under test. Based on the charging environment information and battery attribute information, the predicted characteristics of the battery under test are determined; Based on a preset fusion model, the remaining charging time prediction result of the battery under test is determined according to the prediction characteristics of the battery under test. The step of determining the remaining charging time prediction result of the battery under test based on the prediction characteristics of the battery under test, according to the preset fusion model, includes: Based on the preset first LightGBM model, the first intermediate prediction result is determined according to the prediction characteristics of the battery under test; The first intermediate prediction result is used as the first new feature and added to the prediction feature to obtain the first fused prediction feature; Based on the preset first XGBoost model, and according to the first fusion prediction features, the second intermediate prediction result is determined; The second intermediate prediction result is used as a second new feature and added to the prediction feature to obtain the second fused prediction feature; Based on the preset second LightGBM model, the prediction result of the LightGBM model is determined according to the second fusion prediction feature; Based on the preset second XGBoost model, the XGBoost model prediction result is determined according to the first fusion prediction feature; Based on the preset neural network model, the prediction result of the neural network model is determined according to the prediction features; Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

2. The method according to claim 1, characterized in that, The step of determining the predicted remaining charging time of the battery under test based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model includes: Based on the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined respectively. Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, as well as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

3. The method according to claim 2, characterized in that, The step of determining the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model based on their prediction accuracies includes: Based on several preset test sets, the prediction accuracy of the second LightGBM model, the second XGBoost model, and the neural network model is verified. For any of the aforementioned models, the error rate of the model in each of the aforementioned test sets is determined based on the prediction accuracy verification results corresponding to each of the aforementioned test sets. With the goal of reducing the error rate difference between the various test sets, the error rate of the model in each test set is corrected to obtain the error rate feature set of the model. Based on the error rate feature set of each model, several model weight feature sets are determined; wherein, the model weight feature set includes the error rate features generated by each model on the same test set; Based on the aforementioned set of model weight features, the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model are determined.

4. The method according to claim 3, characterized in that, The step of determining the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model based on the plurality of model weight feature sets includes: Based on the model weight feature set, multiple sets of candidate prediction weights are determined; Based on the aforementioned test sets, the prediction accuracy of each group of candidate prediction weights when applied to the second LightGBM model, the second XGBoost model, and the neural network model is verified. The candidate prediction weights with the highest prediction accuracy are determined as the prediction weights corresponding to the second LightGBM model, the second XGBoost model, and the neural network model.

5. The method according to claim 1, characterized in that, Also includes: Acquire battery charging data from multiple vehicles; wherein, the battery charging data includes charging environment information, battery attribute information, and timestamps corresponding to each information indicator during the battery charging process; The battery charging data is preprocessed to obtain the model training set; Using the model training set, a fusion model is trained to obtain the preset fusion model.

6. The method according to claim 1, characterized in that, The predicted characteristics of the battery under test include at least the vehicle charging method, charging current, charging voltage, ambient temperature, state of charge, and number of battery charge-discharge cycles.

7. A battery remaining charging time prediction device, characterized in that, include: The acquisition module is used to acquire charging environment information and battery attribute information of the battery under test; wherein, the charging environment information represents the power supply conditions of the battery under test and the impact of weather on battery charging, and the battery attribute information represents the state of charge and battery usage of the battery under test. The determination module is used to determine the predicted characteristics of the battery under test based on the charging environment information and battery attribute information. The prediction module is used to determine the remaining charging time prediction result of the battery under test based on the prediction characteristics of the battery under test and a preset fusion model. Specifically, the prediction module is used for: Based on the preset first LightGBM model, the first intermediate prediction result is determined according to the prediction characteristics of the battery under test; The first intermediate prediction result is used as the first new feature and added to the prediction feature to obtain the first fused prediction feature; Based on the preset first XGBoost model, and according to the first fusion prediction features, the second intermediate prediction result is determined; The second intermediate prediction result is used as a second new feature and added to the prediction feature to obtain the second fused prediction feature; Based on the preset second LightGBM model, the prediction result of the LightGBM model is determined according to the second fusion prediction feature; Based on the preset second XGBoost model, the XGBoost model prediction result is determined according to the first fusion prediction feature; Based on the preset neural network model, the prediction result of the neural network model is determined according to the prediction features; Based on the prediction results of the LightGBM model, the XGBoost model, and the neural network model, the remaining charging time prediction result of the battery under test is determined.

8. An electronic device, characterized in that, include: At least one processor and memory; The memory stores computer-executed instructions; The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the method as described in any one of claims 1 to 6.

Images