Method, device and equipment for predicting residual charging time of electric vehicle and medium

Abstract

The invention relates to the technical field of vehicle charging, and discloses a method, a device and equipment for predicting residual charging time of an electric vehicle and a medium. According to the method, the current charging characteristic is constructed by collecting the environment temperature of the electric vehicle, the current electric quantity of a battery, the temperature of a battery cell, the health state of the battery, the current of a charging pile, the power of the charging pile, the charging mode of the current vehicle type and the in-vehicle load use state of the electric vehicle; the influence of external environment factors, charging requirements of different vehicle types and use of loads in the vehicle on the charging rate is integrated. Then, a residual charging time estimation value is obtained through the mapping lookup table, the residual charging time is preliminarily estimated, it is ensured that the residual charging time under the similar charging conditions has consistent preliminary estimation, and the scene adaptability is improved; and the estimated value is corrected and adjusted by using the residual charging time predicted value output by the data driving prediction model, so that the prediction precision of the residual charging time is improved, and the user experience and the intelligent level of charging management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle charging, and in particular to a method, device, equipment and medium for predicting the remaining charging time of an electric vehicle. Background Art

[0002] The electric vehicle remaining charging time estimation technology can predict the time required for the electric vehicle to be fully charged from the current charging state. It is an important function of the electric vehicle charging management system. It can provide users with reference information on the charging progress, making it convenient for users to reasonably arrange travel time and other activities.

[0003] Existing technologies for estimating the remaining charging time of electric vehicles mainly rely on calculation methods based on static parameters such as battery capacity, charging current and current power. Such methods fail to fully consider the various external dynamic factors that affect charging efficiency and charging time, and do not comprehensively consider the charging needs of different models. The estimation accuracy is insufficient, resulting in a deviation between the actual charging time and the predicted value. At the same time, most existing technologies use fixed formulas for calculations and cannot be dynamically updated based on real-time data during the charging process. Therefore, there are obvious deficiencies in the diversification of charging scenarios and real-time adjustments. Summary of the invention

[0004] In view of this, the present invention provides a method, device, equipment and medium for predicting the remaining charging time of an electric vehicle to solve the problem that the prior art does not fully consider the charging requirements of different vehicle models and various external dynamic factors affecting the charging efficiency and charging time, resulting in low accuracy in estimating the remaining charging time.

[0005] In a first aspect, the present invention provides a method for predicting the remaining charging time of an electric vehicle, the method comprising:

[0006] Collecting current charging data of the electric vehicle; wherein the current charging data includes at least some of the following items: ambient temperature, current battery power, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle;

[0007] According to the current charging data, construct the current charging characteristics;

[0008] Using the trained data-driven prediction model, according to the current charging characteristics, a prediction value of the remaining charging time is obtained, and using a preset mapping lookup table, according to the current charging characteristics, an estimated value of the remaining charging time is obtained; wherein the mapping lookup table is used to characterize the mapping relationship between different charging characteristics and the corresponding remaining charging time;

[0009] The target remaining charging time is obtained according to the remaining charging time prediction value and the remaining charging time estimation value.

[0010] Beneficial effects: The present invention collects data such as the ambient temperature of the electric vehicle, the current battery power, the battery cell temperature, the battery health status, the charging pile current, the charging pile power, the charging mode of the current vehicle model, and the in-vehicle load usage status of the electric vehicle to construct the current charging characteristics during charging, and comprehensively considers the influence of external environmental factors, the charging requirements of different vehicle models, and the in-vehicle load usage on the charging rate. Then, the remaining charging time estimate is obtained using a preset mapping lookup table, and a preliminary estimate of the remaining charging time is made to ensure a consistent preliminary estimate of the remaining charging time under similar charging conditions, thereby improving the scene adaptability. In addition, the remaining charging time prediction value output by the data-driven prediction model is used to correct and adjust the remaining charging time estimate, further improving the prediction accuracy of the target remaining charging time, which is conducive to improving the user experience and the level of intelligent charging management.

[0011] In some optional implementations, before obtaining the remaining charging time prediction value according to the current charging characteristics using the trained data-driven prediction model, the method further includes:

[0012] Obtain historical charging data, historical remaining charging time data, and preset model structure;

[0013] An input feature matrix is ​​constructed according to the historical charging data, and a target variable is constructed according to the historical remaining charging time data; wherein the input feature matrix includes historical charging features of multiple charging processes, and the target variable includes multiple historical remaining charging times corresponding to the multiple historical charging features one by one;

[0014] Building a first prediction model based on a preset model structure, training the first prediction model according to an input feature matrix and a target variable, and obtaining a second prediction model;

[0015] The performance evaluation index of the second prediction model is calculated, and the model parameters of the second prediction model are adjusted based on the performance evaluation index to obtain a trained data-driven prediction model.

[0016] Beneficial effects: The prediction model is constructed using the preset model structure, and the historical charging data and the corresponding historical remaining charging time data are used to train the prediction model, so that the prediction model can learn the correlation between different charging conditions and the remaining charging time, and improve the scenario adaptability of the prediction model. The model parameters of the prediction model are adjusted through performance evaluation indicators, and more appropriate model parameter settings are determined to optimize the model and improve the prediction accuracy of the model.

[0017] In some optional implementations, a first prediction model is constructed based on a preset model structure, and the first prediction model is trained according to an input feature matrix and a target variable to obtain a second prediction model, including:

[0018] Constructing a model according to the decision tree model structure to obtain a first prediction model; wherein the first prediction model includes a plurality of decision trees;

[0019] Construct a training set based on the input feature matrix and target variable, and perform random sampling with replacement on the training set to generate multiple sample subsets;

[0020] The sample features in each sample subset are randomly selected, and the corresponding decision tree in the first prediction model is trained using the sample features of each sample subset to obtain the second prediction model.

[0021] Beneficial effects: The present invention uses a decision tree model structure to construct a prediction model for predicting the remaining charging time of an electric vehicle, and uses an input feature matrix constructed based on historical charging data and a target variable constructed based on historical remaining charging time data to train the prediction model. The decision tree is used to learn the remaining charging time under different charging conditions, thereby improving the scenario adaptability of the prediction model.

[0022] In some optional implementations, calculating a performance evaluation index of the second prediction model, and adjusting a model parameter of the second prediction model based on the performance evaluation index to obtain a trained data-driven prediction model includes:

[0023] Calculate the root mean square error and goodness of fit of the second prediction model;

[0024] The model parameters of multiple decision trees in the second prediction model are adjusted according to the root mean square error and the degree of fit to obtain a trained data-driven prediction model; wherein the trained data-driven prediction model includes multiple trained decision trees.

[0025] Beneficial effect: The present invention uses the root mean square error to evaluate the prediction error of the model, and uses the degree of fit to measure the fitting effect of the model. By comprehensively considering these two performance evaluation indicators, a more appropriate model parameter setting is determined, so that the model can achieve a better balance in prediction accuracy and fitting effect, avoiding the occurrence of low prediction accuracy or model overfitting.

[0026] In some optional implementations, using a trained data-driven prediction model to obtain a remaining charging time prediction value according to current charging characteristics, and using a preset mapping lookup table to obtain a remaining charging time estimation value according to current charging characteristics, including:

[0027] Input the current charging characteristics into the trained data-driven prediction model to obtain the prediction results output by multiple trained decision trees respectively;

[0028] The remaining charging time prediction value is calculated based on the average value of multiple prediction results;

[0029] A search is performed in a preset mapping lookup table to obtain an estimated value of the remaining charging time corresponding to the current charging characteristics.

[0030] Beneficial effects: The present invention searches the preset mapping lookup table based on the current charging characteristics, preliminarily determines the estimated value of the remaining charging time under the current charging conditions, ensures that the estimated values ​​of the remaining charging time corresponding to similar charging conditions are basically consistent, and is conducive to improving the scene adaptability. And by integrating the prediction results of multiple decision trees, the overfitting risk of a single decision tree is effectively reduced, and the stability and accuracy of the remaining charging time prediction value are improved.

[0031] In some optional implementations, obtaining a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value includes:

[0032] Determine the target confidence corresponding to the preset mapping lookup table;

[0033] Based on the target confidence, the target remaining charging time is calculated by taking a weighted average of the remaining charging time prediction value and the remaining charging time estimation value.

[0034] Beneficial effect: The present invention performs weighted average on the remaining charging time prediction value and the remaining charging time estimation value, and comprehensively considers the credibility of the preliminary estimation value of the mapping lookup table and the prediction accuracy of the prediction model, which is beneficial to balance the model's adaptability to similar scenarios and prediction accuracy, and improves the remaining charging time prediction accuracy while also improving the calculation efficiency and real-time performance of the remaining charging time.

[0035] In some optional embodiments, the method further comprises:

[0036] The trained data-driven prediction model is updated according to the target remaining charging time and / or the estimated remaining charging time.

[0037] Beneficial effect: By adding the target remaining charging time and / or the remaining charging time estimate to the training process when the model is updated, the feature scenarios are further enriched, and the model is helped to consider more mapping lookup information when dividing the feature scenarios.

[0038] In a second aspect, the present invention provides a device for predicting the remaining charging time of an electric vehicle, the device comprising:

[0039] An acquisition module is used to collect current charging data of the electric vehicle; wherein the current charging data includes at least some of the following items: ambient temperature, current battery power, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle;

[0040] A first processing module, configured to construct a current charging feature according to current charging data;

[0041] The second processing module is used to use the trained data-driven prediction model to obtain a predicted value of the remaining charging time according to the current charging data, and to use a preset mapping lookup table to obtain an estimated value of the remaining charging time according to the current charging data; wherein the mapping lookup table is used to characterize the mapping relationship between different charging characteristics and the corresponding remaining charging time;

[0042] The third processing module is used to obtain the target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value.

[0043] In a third aspect, the present invention provides an electric vehicle, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for predicting the remaining charging time of an electric vehicle according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for predicting the remaining charging time of an electric vehicle according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 is a schematic diagram of a flow chart of a method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention;

[0047] Figure 2 is a flow chart of another method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention;

[0048] Figure 3 is a flow chart of another method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention;

[0049] Figure 4 is a flow chart of another method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention;

[0050] Figure 5is a structural block diagram of a device for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention;

[0051] Figure 6 Schematic diagram of the hardware structure of an electric vehicle according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0053] The existing technology for estimating the remaining charging time of electric vehicles mainly relies on calculation methods based on static parameters such as battery capacity, charging current and current power. The main disadvantage of this type of method is that it fails to fully consider the various external dynamic factors that affect charging efficiency and charging time. For example, changes in ambient temperature will directly affect the charging rate of the battery, and the operation of equipment such as air conditioning in the car will also consume additional power, resulting in a deviation between the actual charging time and the prediction. In addition, the existing technology usually does not consider complex variables such as the status of the charging pile and the different charging requirements of different models, resulting in insufficient estimation accuracy. At the same time, most of the existing technologies use fixed formulas for calculations and cannot be dynamically updated according to real-time data during the charging process. Therefore, there are obvious deficiencies in the diversification of charging scenarios and real-time adjustments. In addition, the existing technical solutions also use cloud data and historical data to improve the accuracy of predictions, but only consider the impact of variables such as current and initial power on charging time, and do not consider a variety of external factors.

[0054] Therefore, the embodiment of the present invention provides a method for predicting the remaining charging time of an electric vehicle. By integrating multi-dimensional data such as battery cell temperature, ambient temperature, in-vehicle load usage, charging pile current / voltage, and vehicle battery status, an accurate data-driven prediction model is trained to achieve dynamic prediction and update of the remaining charging time. The model can adjust the prediction results according to real-time data during the charging process, and adapt to the requirements of different vehicle models, charging piles, and external environmental conditions, providing a more accurate estimation of the remaining charging time, solving the accuracy problems, adaptability problems, and lack of real-time update limitations in the prior art, thereby improving the user experience and the intelligent level of charging management.

[0055] According to an embodiment of the present invention, an embodiment of a method for predicting the remaining charging time of an electric vehicle is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0056] In this embodiment, a method for predicting the remaining charging time of an electric vehicle is provided, which can be used for estimating the remaining charging time of a vehicle, such as an electric vehicle. Figure 1 is a flow chart of a method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0057] Step S101, collecting current charging data of the electric vehicle.

[0058] Specifically, the current charging data includes at least some of the following items: ambient temperature, current battery charge, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle.

[0059] In some optional embodiments, the ambient temperature can be obtained through the temperature sensor of the electric vehicle, battery parameters such as the current battery power (State of Charge, SOC), battery cell temperature, and battery health status (State of Health, SOH) can be obtained through a battery control unit (Battery Management Unit, BMU) that monitors and manages the battery, and charging parameters such as the charging pile current, charging pile power, and charging mode of the current vehicle model can be obtained by viewing the corresponding application.

[0060] It should be noted that the battery management unit (BMU) is a key component used to manage and monitor the performance of battery packs in electric vehicles, energy storage systems and other equipment. It can estimate the SOC by combining the collected voltage, current and other data, and evaluate the battery's SOH by long-term monitoring of the battery's charge and discharge behavior and performance changes. Among them, SOC refers to the percentage of the battery's current remaining power to the total battery capacity, and SOH is used to measure the health of the battery and to reflect the battery's capacity attenuation and internal resistance increase.

[0061] In some optional implementations, the in-vehicle loads of the electric vehicle mainly include air-conditioning systems, ventilation systems, instruments, audio, etc. By obtaining the load usage status such as switch conditions, working gears, etc., the power consumption of the electric vehicle during the charging process is determined, so as to improve the accuracy of the calculation of the remaining charging time.

[0062] Step S102, constructing current charging characteristics according to current charging data.

[0063] Specifically, by performing data preprocessing on the current charging data, such as data cleaning, standardization and screening, and extracting key features from the preprocessed data, the current charging features are constructed. For example, the ambient temperature can be used to obtain environmental features to measure the charging rate under the influence of the external environment; the battery state features can be obtained using battery parameters such as the current battery power, battery cell temperature, and battery health status, and the charging state features can be obtained using charging parameters such as the charging pile current, charging pile power, and the charging mode of the current vehicle model, so as to facilitate the estimation of the remaining charging time corresponding to the current state of the battery, charging pile, and vehicle model; and when predicting the remaining charging time, the in-vehicle load usage is comprehensively considered, the corresponding load usage status features are constructed, and the power loss during the charging process is evaluated.

[0064] Step S103, using the trained data-driven prediction model to obtain a remaining charging time prediction value according to the current charging characteristics, and using a preset mapping lookup table to obtain a remaining charging time estimation value according to the current charging characteristics.

[0065] Specifically, the data-driven prediction model is a model that builds prediction rules based on a large amount of data. It makes predictions by mining patterns, correlations, and regularities in data features. It can use information in current charging characteristics to infer the remaining charging time of electric vehicles and output a predicted value for the remaining charging time.

[0066] Specifically, the mapping lookup table is used to characterize the mapping relationship between different charging characteristics and the corresponding remaining charging time. The mapping lookup table can be preset based on the MAP table, which can store the pre-calculated charging characteristics-remaining charging time mapping, which can be linear or non-linear. When constructing the mapping lookup table, the range and accuracy of the charging characteristics must be determined first, and then the remaining charging time corresponding to different charging characteristics is obtained and stored through theoretical calculations, experimental measurements, statistical induction or computer simulation.

[0067] Step S104, obtaining a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value.

[0068] Specifically, the mapping lookup table is used for query to obtain the estimated value of the remaining charging time, and the known information in the mapping lookup table is used as the prediction basis to ensure a consistent preliminary estimate of the remaining charging time under similar charging conditions, thereby improving the scene adaptability. At the same time, the remaining charging time prediction value output by the data-driven prediction model is used to correct and adjust the preliminary estimate, further improving the prediction accuracy of the remaining charging time and obtaining a more accurate target remaining charging time.

[0069] The method for predicting the remaining charging time of an electric vehicle provided in this embodiment collects data such as the ambient temperature of the electric vehicle, the current battery power, the battery cell temperature, the battery health status, the charging pile current, the charging pile power, the charging mode of the current vehicle model, and the in-vehicle load usage status of the electric vehicle to construct the current charging characteristics during charging, and comprehensively considers the influence of external environmental factors, the charging requirements of different vehicle models, and the in-vehicle load usage on the charging rate. Then, the remaining charging time estimate is obtained using the preset mapping lookup table, and the remaining charging time is preliminarily estimated to ensure that there is a consistent preliminary estimate of the remaining charging time under similar charging conditions, thereby improving the scene adaptability. In addition, the remaining charging time prediction value output by the data-driven prediction model is used to correct and adjust the remaining charging time estimate, further improving the prediction accuracy of the target remaining charging time, which is conducive to improving the user experience and the intelligent level of charging management.

[0070] In this embodiment, a method for predicting the remaining charging time of an electric vehicle is provided, which can be used for estimating the remaining charging time of a vehicle, such as an electric vehicle. Figure 2 is a flow chart of a method for predicting remaining charging time of an electric vehicle according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0071] Step S201, collect the current charging data of the electric vehicle. For details, please refer to Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0072] Step S202: construct current charging characteristics based on current charging data. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0073] Step S203, obtaining historical charging data, historical remaining charging time data and a preset model structure.

[0074] Specifically, the historical charging data includes data items that are consistent with the current charging data, such as the ambient temperature, battery charge, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and the in-vehicle load usage status of the electric vehicle during the historical charging process. The historical remaining charging time data includes the historical remaining charging time corresponding to the historical charging data.

[0075] In some optional implementations, multiple data sources may be integrated in combination with external data sources (such as weather forecasts, charging pile network data, etc.) to further improve the accuracy of charging time prediction and the adaptability of the model.

[0076] In some optional embodiments, the preset model structure can be a decision tree model structure and other machine learning model structures, such as a support vector machine model (SVM), a deep learning model (such as a convolutional neural network), which can be selected and adjusted according to actual application requirements, but the present invention is not limited to this.

[0077] Step S204, constructing an input feature matrix based on historical charging data, and constructing a target variable based on historical remaining charging time data.

[0078] Specifically, the input feature matrix includes multiple historical charging features of the charging process, and the target variable includes multiple historical remaining charging times corresponding to the multiple historical charging features. Exemplarily, the input feature matrix is ​​set to X = [x1, x2, ..., x n ], where x i represents the historical charging characteristics of the i-th charging process; the target variable is set to y = [y1, y2, ..., y n ], where y i Represents the historical remaining charging time of the i-th charging process, the historical remaining charging time y i With historical charging characteristics x i One to one correspondence.

[0079] Step S205, constructing a first prediction model based on a preset model structure, training the first prediction model according to the input feature matrix and the target variable to obtain a second prediction model.

[0080] In some optional implementations, the above step S205 includes the following steps:

[0081] Step a1, constructing a model according to the decision tree model structure to obtain a first prediction model; wherein the first prediction model includes multiple decision trees.

[0082] Specifically, a random forest (RF) model is constructed based on the decision tree model structure to obtain the first prediction model. It should be noted that the core idea of ​​the random forest is to construct multiple decision trees, each tree is trained according to a different feature subset, and finally predict by integrating the results of all decision trees. The specific construction process can refer to the detailed description of the relevant technology, which will not be repeated here.

[0083] Step a2: construct a training set based on the input feature matrix and the target variable, and perform random sampling with replacement on the training set to generate multiple sample subsets.

[0084] Specifically, multiple sample subsets are generated from the training data set by random sampling with replacement. Assume that the original training set is D = {(X1, y1), (X2, y2), …, (X n ,y n )}, each time randomly sampling m samples from D to generate a sample subset Dj, the sample subset Dj is as follows:

[0085] D j ={(X j1 ,y j1 ),(X j2 ,y j2 ),…,(X jm ,y jm )}

[0086] Step a3: randomly select sample features from each sample subset, and use the sample features of each sample subset to train the corresponding decision tree in the first prediction model to obtain the second prediction model.

[0087] Specifically, for each sample subset Dj, some features are randomly selected for training to build a decision tree. The decision tree recursively divides the feature space and selects the best split point to maximize the purity of the target variable. The splitting standard usually uses information gain or Gini Index. In this embodiment, for the regression task of remaining charging time prediction, the most commonly used standard is to minimize the mean square error (MSE):

[0088]

[0089] Where N is the number of samples, y i is the actual historical remaining charging time, is the estimated remaining charging time.

[0090] Furthermore, after the decision tree is divided into leaf nodes, each leaf node corresponds to a sample subset of charging data. For regression problems, the output of the leaf node is the average value of all sample features in the node for the remaining charging time:

[0091]

[0092] Among them, D leaf is the sample feature set in the leaf node, |D leaf | is the number of samples in the sample feature set, y i is the real historical remaining charging time corresponding to the corresponding sample feature.

[0093] Furthermore, by repeating the above steps, k decision trees are generated to obtain a second prediction model.

[0094] The present invention uses a decision tree model structure to construct a prediction model for predicting the remaining charging time of an electric vehicle, and uses an input feature matrix constructed based on historical charging data and a target variable constructed based on historical remaining charging time data to train the prediction model. The decision tree is used to learn the remaining charging time under different charging conditions, thereby improving the scenario adaptability of the prediction model.

[0095] Step S206, calculating the performance evaluation index of the second prediction model, and adjusting the model parameters of the second prediction model based on the performance evaluation index to obtain a trained data-driven prediction model.

[0096] In some optional implementations, the above step S206 includes:

[0097] Step b1, calculating the root mean square error and the goodness of fit of the second prediction model.

[0098] Exemplarily, in order to evaluate the performance of the second prediction model, i.e., the random forest model, the root mean square error (RMSE) can be used to evaluate the prediction error size of the model:

[0099]

[0100] For example, R 2 The coefficient of determination measures the goodness of fit of the model:

[0101]

[0102] Step b2, adjusting the model parameters of multiple decision trees in the second prediction model according to the root mean square error and the goodness of fit to obtain a trained data-driven prediction model; wherein the trained data-driven prediction model includes multiple trained decision trees.

[0103] Specifically, the accuracy of the root mean square error and the goodness of fit can be set respectively. As long as the root mean square error and the goodness of fit of the second prediction model have not reached the set accuracy, the decision tree training process is repeated, and the model parameters of multiple decision trees are continuously adjusted until convergence to obtain a trained data-driven prediction model.

[0104] The present invention uses the root mean square error to evaluate the prediction error of the model, and uses the degree of fit to measure the fitting effect of the model. By comprehensively considering these two performance evaluation indicators, a more appropriate model parameter setting is determined, so that the model can achieve a better balance in prediction accuracy and fitting effect, avoiding low prediction accuracy or model overfitting.

[0105] In the embodiment of the present invention, a prediction model is constructed using a preset model structure, and the prediction model is trained using historical charging data and corresponding historical remaining charging time data, so that the prediction model learns the correlation between different charging conditions and remaining charging time, thereby improving the scenario adaptability of the prediction model. The model parameters of the prediction model are adjusted through performance evaluation indicators, a more appropriate model parameter setting is determined, the model is optimized, and the prediction accuracy of the model is improved.

[0106] Step S207, using the trained data-driven prediction model to obtain a predicted value of the remaining charging time according to the current charging characteristics, and using a preset mapping lookup table to obtain an estimated value of the remaining charging time according to the current charging characteristics.

[0107] Specifically, the above step S207 includes:

[0108] Step S2071, input the current charging characteristics into the trained data-driven prediction model to obtain prediction results output by multiple trained decision trees respectively.

[0109] Specifically, for the newly input current charging characteristic X new , each decision tree Tj will give a prediction result

[0110] Step S2072: Calculate the remaining charging time prediction value based on the average value of multiple prediction results.

[0111] Specifically, the final remaining charging time prediction value is the average of the predictions of all decision trees:

[0112]

[0113] Where k is the number of decision trees, is the prediction result of the jth decision tree.

[0114] Step S2073, searching in a preset mapping lookup table to obtain an estimated value of the remaining charging time corresponding to the current charging characteristics.

[0115] For example, an input feature in the MAP table may be:

[0116] MAP time =f(SOC,Power)

[0117] Among them, function f is used to calculate the estimated value of the remaining charging time in the charging MAP table under specific SOC and charging pile power conditions.

[0118] The present invention searches the preset mapping lookup table based on the current charging characteristics, preliminarily determines the estimated value of the remaining charging time under the current charging conditions, ensures that the estimated values ​​of the remaining charging time corresponding to similar charging conditions are basically consistent, and is conducive to improving the scene adaptability. And integrates the prediction results of multiple decision trees, effectively reduces the overfitting risk of a single decision tree, and improves the stability and accuracy of the remaining charging time prediction value.

[0119] Step S208, obtaining a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value.

[0120] Specifically, the above step S208 includes:

[0121] Step S2081, determining the target confidence corresponding to the preset mapping lookup table.

[0122] Specifically, the target confidence α corresponding to the mapping lookup table is used as an adjustment parameter of the weighted average to control the balance between the MAP table estimation and the model prediction.

[0123] Step S2082: Based on the target confidence, a weighted average is taken according to the predicted value of the remaining charging time and the estimated value of the remaining charging time to calculate the target remaining charging time.

[0124] Specifically, in the final prediction stage, the remaining charging time prediction value of the random forest can be weighted averaged according to the remaining charging time estimation value recorded in the MAP table. If the information in the MAP table is more reliable, it can be given a larger weight:

[0125]

[0126] in, is the estimated remaining charging time from the MAP table, is the remaining charging time prediction value of random forest, and α is the parameter for adjusting the weight.

[0127] The present invention performs weighted averaging on the predicted value of the remaining charging time and the estimated value of the remaining charging time, and comprehensively considers the credibility of the preliminary estimated value of the mapping lookup table and the prediction accuracy of the prediction model, which is beneficial to balancing the model's adaptability to similar scenarios and the prediction accuracy, and improves the remaining charging time prediction accuracy while also improving the calculation efficiency and real-time performance of the remaining charging time.

[0128] In some optional implementations, the trained data-driven prediction model may also be updated according to the target remaining charging time and / or the estimated remaining charging time.

[0129] For example, the MAP table estimated time corresponding to the current battery power and the charging pile power is added as a new feature to the input feature matrix X, which helps the decision tree to consider more MAP table information when dividing, thereby improving the prediction accuracy. The target remaining charging time determined according to the current charging characteristics can also be added to the input feature matrix X to further enrich the feature scenarios.

[0130] The remaining charging time prediction scheme for an electric vehicle of the present invention is described in detail below with reference to a specific application example.

[0131] like Figure 3 As shown, this application example proposes a method for predicting the remaining charging time of electric vehicles based on the random forest algorithm, which comprehensively considers multiple factors such as battery cell temperature, SOC, current state of the battery, ambient temperature, charging current and charging pile information, and uses the random forest algorithm to improve the prediction accuracy.

[0132] Random forest is an integrated learning algorithm that builds multiple decision tree models and votes or averages their prediction results to output the final prediction result. Random forest is very effective in dealing with nonlinear relationships and complex interactions between features, and has strong resistance to overfitting, making it very suitable for complex electric vehicle charging remaining time prediction problems.

[0133] Combine the following Figure 4 Detailed description of the remaining charging time estimation process based on the random forest algorithm, combined with Figure 4 As shown, the process mainly includes the following steps:

[0134] Step 1: Data preparation and feature selection.

[0135] First, collect relevant data during the electric vehicle charging process. These data will serve as input features of the model, including but not limited to: current battery power, ambient temperature, charging pile power, charging mode, battery health status, in-car air conditioning usage, and historical remaining charging time data.

[0136] Secondly, set the input feature matrix to be X = [x1, x2, ..., x n ], where x i represents the historical charging characteristics of the i-th charging process; the target variable is set to y = [y1, y2, ..., y n ], where y i Represents the historical remaining charging time of the i-th charging process, the historical remaining charging time y i With historical charging characteristics x i One to one correspondence.

[0137] Step 2: Random forest model construction.

[0138] Step 2.1, Bootstrap sampling: Generate multiple sample subsets by random sampling with replacement from the training data set. Assume that the original training set is D = {(X1, y1), (X2, y2), …, (X n ,y n )}, each time randomly sampling m samples from D to generate a sample subset Dj, the sample subset Dj is as follows:

[0139] D j ={(X j1 ,y j1 ),(X j2 ,y j2 ),…,(X jm ,y jm )}

[0140] Step 2.2, build a decision tree: For each sample subset Dj, randomly select some features for training and build a decision tree. The decision tree recursively divides the feature space and selects the best split point to maximize the purity of the target variable. The splitting standard usually uses information gain or Gini Index. In this embodiment, for the regression task of remaining charging time prediction, the most commonly used standard is to minimize the mean square error (MSE):

[0141]

[0142] Where N is the number of samples, y i is the actual historical remaining charging time, is the estimated remaining charging time.

[0143] Step 2.3, prediction on leaf nodes: After the decision tree is divided into leaf nodes, each leaf node corresponds to a sample subset of charging data. For regression problems, the output of the leaf node is the average value of all sample features in the node for the remaining charging time:

[0144]

[0145] Among them, D leaf is the sample feature set in the leaf node, |D leaf | is the number of samples in the sample feature set, y i is the real historical remaining charging time corresponding to the corresponding sample feature.

[0146] Step 2.4, by repeating the above steps, k decision trees are generated to obtain a random forest model.

[0147] Step 3, random forest prediction.

[0148] For the new input data Xnew , each decision tree Tj will give a prediction result Final remaining charging time prediction is the average of the predictions of all decision trees:

[0149]

[0150] Where k is the number of decision trees, is the prediction result of the jth decision tree. By integrating the predictions of multiple trees, random forests can effectively reduce the overfitting risk of a single tree and improve the stability and accuracy of the prediction.

[0151] Step 4: Apply the charging MAP table information.

[0152] When implementing the random forest algorithm, the prior information provided by the charging MAP table can be used to improve the model. The specific steps are as follows:

[0153] Step 4.1, MAP table features as input: add the MAP table estimated time corresponding to the current battery power and charging pile power as a new feature to the input feature matrix X, to help the decision tree consider more MAP table information when dividing, thereby improving the prediction accuracy. For example: an input feature in the MAP table can be:

[0154] MAP rime =f(SOC,Power)

[0155] Among them, function f is used to calculate the estimated value of the remaining charging time in the charging MAP table under specific SOC and charging pile power conditions.

[0156] Step 4.2, weight adjustment: In the final prediction stage, the remaining charging time prediction value of the random forest can be weighted averaged according to the remaining charging time estimation value recorded in the MAP table. If the information in the MAP table is more reliable, it can be given a larger weight:

[0157]

[0158] in, is the estimated remaining charging time from the MAP table, is the remaining charging time prediction value of random forest, and α is the parameter for adjusting the weight.

[0159] Step 5: Model performance evaluation.

[0160] To evaluate the performance of the random forest model, we can use the root mean square error to evaluate the prediction error of the model and use R 2 The coefficient of determination measures the goodness of fit of the model.

[0161] The above application examples combine regularized data such as charging MAP tables, and use this known information as the basis of the model to ensure consistent preliminary estimates under similar charging conditions. Through data-driven algorithms such as random forests, the prediction is dynamically adjusted based on historical charging data, ambient temperature, SOC and other dynamic variables to further improve the prediction accuracy. In addition, multiple decision trees are integrated, which have strong noise resistance and adaptability to different charging scenarios, and have high robustness and accuracy. Finally, the rule-driven MAP table estimate and the data-driven model prediction are combined to use the advantages of both to reduce errors, adapt to diverse charging scenarios, and further improve prediction accuracy.

[0162] In this embodiment, a device for predicting the remaining charging time of an electric vehicle is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0163] This embodiment provides a device for predicting the remaining charging time of an electric vehicle. Figure 5 As shown, including:

[0164] The acquisition module 501 is used to collect the current charging data of the electric vehicle; wherein the current charging data includes at least some of the following items: ambient temperature, current battery power, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle;

[0165] A first processing module 502 is used to construct a current charging feature according to the current charging data;

[0166] The second processing module 503 is used to use the trained data-driven prediction model to obtain a predicted value of the remaining charging time according to the current charging data, and to use a preset mapping lookup table to obtain an estimated value of the remaining charging time according to the current charging data; wherein the mapping lookup table is used to characterize the mapping relationship between different charging characteristics and the corresponding remaining charging time;

[0167] The third processing module 504 is used to obtain a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value.

[0168] In some optional implementations, before obtaining a predicted value of the remaining charging time according to current charging characteristics using a trained data-driven prediction model, the device is further used to:

[0169] Obtain historical charging data, historical remaining charging time data, and preset model structure;

[0170] An input feature matrix is ​​constructed according to the historical charging data, and a target variable is constructed according to the historical remaining charging time data; wherein the input feature matrix includes historical charging features of multiple charging processes, and the target variable includes multiple historical remaining charging times corresponding to the multiple historical charging features one by one;

[0171] Building a first prediction model based on a preset model structure, training the first prediction model according to an input feature matrix and a target variable, and obtaining a second prediction model;

[0172] The performance evaluation index of the second prediction model is calculated, and the model parameters of the second prediction model are adjusted based on the performance evaluation index to obtain a trained data-driven prediction model.

[0173] In some optional embodiments, the device is also used for:

[0174] Constructing a model according to the decision tree model structure to obtain a first prediction model; wherein the first prediction model includes a plurality of decision trees;

[0175] Construct a training set based on the input feature matrix and target variable, and perform random sampling with replacement on the training set to generate multiple sample subsets;

[0176] The sample features in each sample subset are randomly selected, and the corresponding decision tree in the first prediction model is trained using the sample features of each sample subset to obtain the second prediction model.

[0177] In some optional embodiments, the device is also used for:

[0178] Calculate the root mean square error and goodness of fit of the second prediction model;

[0179] The model parameters of multiple decision trees in the second prediction model are adjusted according to the root mean square error and the degree of fit to obtain a trained data-driven prediction model; wherein the trained data-driven prediction model includes multiple trained decision trees.

[0180] In some optional implementations, the second processing module 503 is further configured to:

[0181] Input the current charging characteristics into the trained data-driven prediction model to obtain the prediction results output by multiple trained decision trees respectively;

[0182] The remaining charging time prediction value is calculated based on the average value of multiple prediction results;

[0183] A search is performed in a preset mapping lookup table to obtain an estimated value of the remaining charging time corresponding to the current charging characteristics.

[0184] In some optional implementations, the third processing module 504 is further configured to:

[0185] Determine the target confidence corresponding to the preset mapping lookup table;

[0186] Based on the target confidence, the target remaining charging time is calculated by taking a weighted average of the remaining charging time prediction value and the remaining charging time estimation value.

[0187] In some optional embodiments, the device is also used for:

[0188] The trained data-driven prediction model is updated according to the target remaining charging time and / or the estimated remaining charging time.

[0189] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0190] The electric vehicle remaining charging time prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0191] The embodiment of the present invention also provides an electric vehicle having the above Figure 5 The electric vehicle remaining charging time prediction device shown in the figure can be deployed in the vehicle-mounted system of the electric vehicle, or implemented through a cloud computing platform, data can be uploaded to the cloud for centralized analysis, and the results can be provided to users through mobile devices, thereby improving the flexibility of hardware deployment.

[0192] See also Figure 6 , Figure 6 is a schematic diagram of the structure of an electric vehicle provided by an optional embodiment of the present invention, such as Figure 6As shown, the electric vehicle includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses for communication, and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electric vehicle, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0193] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0194] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0195] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electric vehicle, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the electric vehicle via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0196] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0197] The electric vehicle further comprises a communication interface 30 for the electric vehicle to communicate with other devices or a communication network.

[0198] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0199] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0200] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for predicting the remaining charging time of an electric vehicle, characterized in that: The method comprises: Collecting current charging data of the electric vehicle; wherein the current charging data includes at least some of the following items: ambient temperature, current battery power, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle; constructing a current charging feature according to the current charging data; Using the trained data-driven prediction model, according to the current charging characteristics, a prediction value of the remaining charging time is obtained, and using a preset mapping lookup table, according to the current charging characteristics, an estimated value of the remaining charging time is obtained; wherein the mapping lookup table is used to characterize a mapping relationship between different charging characteristics and corresponding remaining charging times; A target remaining charging time is obtained according to the remaining charging time prediction value and the remaining charging time estimation value.

2. The method according to claim 1, characterized in that Before obtaining a remaining charging time prediction value according to the current charging characteristics using the trained data-driven prediction model, the method further includes: Obtain historical charging data, historical remaining charging time data, and preset model structure; An input feature matrix is ​​constructed according to the historical charging data, and a target variable is constructed according to the historical remaining charging time data; wherein the input feature matrix includes a plurality of historical charging features of the charging process, and the target variable includes a plurality of historical remaining charging times corresponding to the plurality of historical charging features one by one; Building a first prediction model based on the preset model structure, and training the first prediction model according to the input feature matrix and the target variable to obtain a second prediction model; A performance evaluation index of the second prediction model is calculated, and model parameters of the second prediction model are adjusted based on the performance evaluation index to obtain a trained data-driven prediction model.

3. The method according to claim 2, characterized in that The step of constructing a first prediction model based on the preset model structure and training the first prediction model according to the input feature matrix and the target variable to obtain a second prediction model includes: Construct a model according to the decision tree model structure to obtain a first prediction model; wherein the first prediction model includes multiple decision trees; Constructing a training set according to the input feature matrix and the target variable, and performing random sampling with replacement on the training set to generate multiple sample subsets; The sample features in each sample subset are randomly selected, and the corresponding decision tree in the first prediction model is trained using the sample features of each sample subset to obtain the second prediction model.

4. The method according to claim 3, characterized in that The calculating the performance evaluation index of the second prediction model and adjusting the model parameters of the second prediction model based on the performance evaluation index to obtain a trained data-driven prediction model includes: Calculating the root mean square error and the goodness of fit of the second prediction model; The model parameters of multiple decision trees in the second prediction model are adjusted according to the root mean square error and the degree of fit to obtain a trained data-driven prediction model; wherein the trained data-driven prediction model includes multiple trained decision trees.

5. The method according to claim 4, characterized in that The method of using the trained data-driven prediction model to obtain a predicted value of the remaining charging time according to the current charging characteristics, and using a preset mapping lookup table to obtain an estimated value of the remaining charging time according to the current charging characteristics, includes: Inputting the current charging characteristics into a trained data-driven prediction model to obtain prediction results output by a plurality of trained decision trees respectively; The remaining charging time prediction value is calculated based on the average value of multiple prediction results; A search is performed in a preset mapping lookup table to obtain an estimated value of the remaining charging time corresponding to the current charging characteristic.

6. The method according to any one of claims 1 to 5, characterized in that The obtaining a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value includes: Determine the target confidence corresponding to the preset mapping lookup table; Based on the target confidence, and by taking a weighted average of the remaining charging time prediction value and the remaining charging time estimation value, a target remaining charging time is calculated.

7. The method according to any one of claims 1 to 5, characterized in that The method further comprises: The trained data-driven prediction model is updated according to the target remaining charging time and / or the estimated remaining charging time.

8. A device for predicting the remaining charging time of an electric vehicle, characterized in that: The device comprises: An acquisition module is used to collect current charging data of the electric vehicle; wherein the current charging data includes at least some of the following items: ambient temperature, current battery power, battery cell temperature, battery health status, charging pile current, charging pile power, charging mode of the current vehicle model, and in-vehicle load usage status of the electric vehicle; A first processing module, configured to construct a current charging feature according to the current charging data; A second processing module is used to use the trained data-driven prediction model to obtain a remaining charging time prediction value according to the current charging characteristics, and to use a preset mapping lookup table to obtain a remaining charging time estimation value according to the current charging characteristics; wherein the mapping lookup table is used to characterize a mapping relationship between different charging characteristics and corresponding remaining charging times; The third processing module is used to obtain a target remaining charging time according to the remaining charging time prediction value and the remaining charging time estimation value.

9. An electric vehicle, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for predicting the remaining charging time of an electric vehicle according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method for predicting the remaining charging time of an electric vehicle according to any one of claims 1 to 7.

Images