Battery health state prediction method and device, equipment and storage medium

Abstract

The invention relates to the technical field of data processing, and discloses a battery health state prediction method, device and equipment and a storage medium, and the method comprises the steps: obtaining historical charging data of a target vehicle and historical driving data corresponding to the historical charging data; analyzing the historical charging data to obtain historical charging fragment data; performing feature extraction on the historical charging fragment data to obtain attribute features of a battery in the target vehicle; performing feature analysis based on the historical driving data and the attribute features to obtain fusion features; and predicting the state of the battery based on the fusion features to obtain a state prediction result of the battery. According to the method, the battery state can be accurately predicted in time.

Description

Technical Field

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

[0002] As an important power source of new energy vehicles, the battery plays a crucial role during the driving process of the vehicle. Therefore, to ensure the normal driving of the vehicle, it is necessary to accurately predict the battery state in a timely manner. Summary of the Invention

[0003] In view of this, the present invention provides a method, device, equipment and storage medium for predicting the state of health of a battery to solve the problem of accurately predicting the battery state.

[0004] In a first aspect, the present invention provides a method for predicting the state of health of a battery, the method comprising:

[0005] Obtaining historical charging data of a target vehicle and historical driving data corresponding to the historical charging data;

[0006] Analyzing the historical charging data to obtain historical charging segment data;

[0007] Extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle;

[0008] Performing feature analysis based on the historical driving data and the attribute features to obtain a fusion feature;

[0009] Predicting the state of the battery based on the fusion feature to obtain a state prediction result of the battery.

[0010] The method for predicting the state of health of a battery provided by the embodiments of the present invention obtains historical charging data of a target vehicle and historical driving data corresponding to the historical charging data, analyzes the historical charging data to obtain historical charging segment data that can balance feature effectiveness and availability; extracts features from the historical charging segment data to obtain attribute features of the battery in the target vehicle to improve the accuracy of predicting the state of health of the battery; performs feature analysis based on the historical driving data and the attribute features to obtain a fusion feature, thereby further improving the accuracy of predicting the state of health of the battery; predicts the state of the battery based on the fusion feature to obtain a state prediction result of the battery, so as to facilitate timely reminding the user to perform battery maintenance and replacement based on the state prediction result.

[0011] In some optional embodiments, the extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle includes:

[0012] Perform the first voltage sampling of the battery cells on the historical charging segment data to obtain the cell voltage sequences of the individual battery cells in the battery;

[0013] Concatenate the cell voltage sequences to obtain the voltage feature of the battery, and the attribute feature includes the voltage feature.

[0014] The battery health state prediction method provided by the embodiments of the present invention performs the first voltage sampling of the battery cells on the historical charging segment data to obtain the cell voltage sequences of the individual battery cells in the battery, thereby obtaining the voltage feature of the battery, providing a basis for the determination of the fusion feature and the determination of the battery health state.

[0015] In some alternative embodiments, the extracting the attribute feature of the battery in the target vehicle from the historical charging segment data includes:

[0016] Perform the second voltage sampling of the battery cells on the historical charging segment data to obtain the highest cell voltage of all the battery cells at each sampling moment, and form the highest cell voltage sequence;

[0017] Obtain the corresponding highest cell current sequence based on the highest cell voltage sequence to obtain the power increment at each sampling moment;

[0018] Perform interpolation based on the power increments at each sampling moment to obtain the power increments corresponding to each voltage point in the historical charging segment data, and form the power increment sequence, and the attribute feature includes the power increment sequence.

[0019] The battery health state prediction method provided by the embodiments of the present invention performs the second voltage sampling of the battery cells on the historical charging segment data to obtain the highest cell voltage of all the battery cells at each sampling moment, and form the highest cell voltage sequence to obtain the corresponding highest cell current sequence; obtain the corresponding highest cell current sequence based on the highest cell voltage sequence to calculate the power increment at each sampling moment; perform interpolation based on the power increments at each sampling moment to obtain the power increments corresponding to each voltage point in the historical charging segment data, and form the power increment sequence, providing a basis for the determination of the fusion feature and the determination of the battery health state.

[0020] In some alternative embodiments, the extracting the attribute feature of the battery in the target vehicle from the historical charging segment data includes:

[0021] Obtain the temperature of the battery at each sampling moment to obtain the temperature sequence, and the attribute feature includes the temperature sequence.

[0022] In some alternative embodiments, the attribute features include the voltage feature of the battery, the power increment sequence, and the temperature sequence. The feature analysis based on the historical driving data and the attribute features to obtain the fusion feature of the battery includes:

[0023] Perform statistical analysis on the voltage feature, the power increment sequence, and the temperature sequence respectively to obtain corresponding statistical analysis features;

[0024] Based on the fusion of the statistical analysis features and the historical driving data, obtain the fusion feature.

[0025] The battery health state prediction method provided by the embodiments of the present invention performs statistical analysis on the voltage feature, the power increment sequence, and the temperature sequence respectively to obtain corresponding statistical analysis features, so as to fuse the statistical analysis features with the historical driving data to obtain the fusion feature, and predict the state of the battery based on the fusion feature, and improve the accuracy of predicting the battery health state.

[0026] In some alternative embodiments, the predicting the state of the battery based on the fusion feature to obtain the state prediction result of the battery includes:

[0027] Input the fusion feature into the first state prediction model to obtain a first state prediction result;

[0028] Determine the state prediction result of the battery based on the first state prediction result.

[0029] The battery health state prediction method provided by the embodiments of the present invention inputs the fusion feature into the first state prediction model to obtain a first state prediction result, so as to improve the accuracy of predicting the battery health state.

[0030] In some alternative embodiments, the attribute features include the voltage feature of the battery and the time feature sequence, and the time feature sequence includes the power increment sequence and the temperature sequence. The determining the state prediction result of the battery based on the first state prediction result includes:

[0031] Obtain the fourth feature in the first state prediction result;

[0032] Input the first feature, the second feature, the third feature, and the fourth feature into the prediction unit of the second state prediction model to obtain a second state prediction result; wherein, the structural complexity of the second state prediction model is greater than that of the first state prediction model;

[0033] Determine the state prediction result of the battery based on the second state prediction result.

[0034] The battery health state prediction method provided by the embodiment of the present invention obtains the fourth feature of the first state prediction result, inputs the voltage feature, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model, and obtains the corresponding first feature, second feature, and third feature, so as to further improve the dimension and accuracy of the extraction of battery state-related features; and performs state prediction on the concatenation result of the first feature, second feature, third feature, and fourth feature through the prediction unit based on the second state prediction model to obtain the second state prediction result, so as to further improve the accuracy of the battery health state prediction.

[0035] In some optional embodiments, the first state prediction model is constructed based on a machine learning model, and the second state prediction model is constructed based on a deep learning neural network.

[0036] In some optional embodiments, the attribute features include the voltage feature of the battery and the time feature sequence, the time feature sequence includes the charge increment sequence and the temperature sequence, and predicting the state of the battery based on the fusion feature to obtain the state prediction result of the battery includes:

[0037] Input the voltage feature, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model to obtain the corresponding first feature, second feature, and third feature;

[0038] Input the first feature, the second feature, and the third feature into the prediction unit of the second state prediction model to obtain the second state prediction result;

[0039] Determine the state prediction result of the battery based on the second state prediction result.

[0040] The battery health state prediction method provided by the embodiment of the present invention inputs the voltage feature, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model to obtain the corresponding first feature, second feature, and third feature, so as to further improve the dimension and accuracy of the extraction of battery state-related features; and fuses the first feature, second feature, and third feature through the prediction unit based on the second state prediction model to obtain the second state prediction result, so as to improve the accuracy of the battery health state prediction.

[0041] In a second aspect, the present invention provides a battery health state prediction device, and the device includes:

[0042] A data acquisition module for acquiring the historical charging data of the target vehicle and the historical driving data corresponding to the historical charging data;

[0043] A data analysis module for analyzing the historical charging data to obtain historical charging segment data;

[0044] A feature extraction module for extracting features from the historical charging segment data to obtain the attribute features of the battery in the target vehicle;

[0045] A feature analysis module for performing feature analysis based on the historical driving data and the attribute features to obtain fused features;

[0046] A state prediction module for predicting the state of the battery based on the fused features to obtain the state prediction result of the battery.

[0047] The battery health state prediction device provided by the embodiment of the present invention obtains the historical charging data of the target vehicle and the corresponding historical driving data, analyzes the historical charging data, and then obtains historical charging segment data that can balance the effectiveness and availability of features; extracts features from the historical charging segment data to obtain the attribute features of the battery in the target vehicle, so as to improve the accuracy of predicting the battery health state; performs feature analysis based on the historical driving data and the attribute features to obtain fused features, thereby further improving the accuracy of predicting the battery health state; predicts the state of the battery based on the fused features to obtain the state prediction result of the battery, which is convenient for timely reminding the user to perform battery maintenance and replacement based on the state prediction result.

[0048] In a third aspect, the present invention provides a computer device, including:

[0049] A memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the battery health state prediction method according to the first aspect or any corresponding embodiment thereof;

[0050] In a fourth aspect, the present invention provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to cause a computer to execute the battery health state prediction method according to the first aspect or any corresponding embodiment thereof.

[0051] It should be noted that for the corresponding beneficial effects of the battery health state prediction device, computer device, and computer-readable storage medium provided by the embodiments of the present invention, please refer to the description of the corresponding beneficial effects of the battery health state prediction method above, and details are not described herein again.

[0052] The beneficial effects of the present invention:

[0053] (1) By performing data analysis on historical charging data, historical charging segment data that can balance the effectiveness and accessibility of features is obtained, while reducing the difficulty of feature extraction and lowering the time and economic costs of predicting the battery health state, thus facilitating large-scale applications.

[0054] (2) By inputting the fusion features obtained through feature analysis based on historical driving data and attribute features into the first state prediction model to obtain the first state prediction result, the accuracy of predicting the battery health state is improved.

[0055] (3) By performing feature extraction and fusion on attribute features and fusion features to obtain the second state prediction result, the accuracy of predicting the battery health state is improved.

[0056] (4) By fusing the first state prediction result with the second state prediction result obtained through feature extraction and fusion based on attribute features and fusion features again, the accuracy and timeliness of predicting the battery state are further improved.

[0057] (5) By predicting the battery health state based on multi-dimensional features such as the first feature, the second feature, the third feature, and the fourth feature, compared with the prediction of the battery health state based on a single dimension, the authenticity and reliability of predicting the battery health state can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0059] Figure 1 is a flowchart of the method for predicting the battery health state according to an embodiment of the present invention;

[0060] Figure 2 is a diagram of the OCV-SOC correspondence according to an embodiment of the present invention;

[0061] Figure 3 is a diagram of the user charging behavior according to an embodiment of the present invention;

[0062] Figure 4 is a historical charging segment according to an embodiment of the present invention;

[0063] Figure 5 is a matrix of voltage features according to an embodiment of the present invention;

[0064] Figure 6It is another process schematic diagram of the battery health state prediction method according to an embodiment of the present invention;

[0065] Figure 7 It is a data flow diagram of the battery health state prediction according to an embodiment of the present invention;

[0066] Figure 8 It is a structural block diagram of the battery health state prediction device according to an embodiment of the present invention

[0067] Figure 9 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed implementation manners

[0068] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0069] In the related technologies of battery health state prediction, when a user uses a computer device to predict the battery health state, it is usually necessary to perform feature extraction through an electrochemical impedance spectroscopy test method. However, in the actual environment, the battery system sensor has a large noise, and a specific instrument is required to perform the high-frequency electrochemical impedance spectroscopy test working condition, which is not conducive to large-scale application. Moreover, during the test process, it is necessary to unify the length of the lithium battery cycle data. In the actual use working condition, the starting point, duration, and charge and discharge depth of the charge and discharge behavior of the lithium battery are random, resulting in the difficulty of actual application of this method.

[0070] Based on this, the embodiments of the present invention provide a battery health state prediction method, device, computer device and storage medium. By obtaining the historical charging data of a target vehicle and the historical driving data corresponding to the historical charging data, the data analysis of the historical charging data is performed, and then the historical charging segment data that can balance the feature effectiveness and availability is obtained; by performing feature extraction on the historical charging segment data, the attribute features of the battery in the target vehicle are obtained to improve the accuracy of predicting the battery health state; by performing feature analysis based on the historical driving data and the attribute features, the fusion features are obtained, thereby further improving the accuracy of predicting the battery health state; by predicting the state of the battery based on the fusion features, the state prediction result of the battery is obtained, so as to facilitate timely reminding the user to perform battery maintenance and replacement based on the state prediction result.

[0071] In this embodiment, a battery health state prediction method is provided, Figure 1is a flowchart of a method for predicting the state of health of a battery according to an embodiment of the present invention. As Figure 1 shown, the process includes the following steps:

[0072] Step S101, obtain the historical charging data of the target vehicle and the historical driving data corresponding to the historical charging data.

[0073] Among them, the historical driving data is the data that the target vehicle has traveled when it stops driving and charges. That is, the historical driving data is data with the same time series as the historical charging data. The historical charging data and the charging data can also be stored in the database corresponding to the target vehicle driving data in a mapped manner, and can also be stored in the cloud server in a key-value manner.

[0074] In some optional embodiments, the historical charging data of the target vehicle may include the charging voltage, charging current, and charging temperature of each battery cell during charging. The historical driving data of the target vehicle may include the cumulative driving mileage and the service life of the target vehicle. Among them, the specific data content of the historical charging data and the historical driving data of the target vehicle can also be increased or decreased based on actual needs, and no limitation is made here.

[0075] Step S102, analyze the historical charging data to obtain historical charging segment data.

[0076] Among them, when analyzing the historical charging data of the target vehicle, the historical data can be cleaned and screened to obtain the historical charging data segment corresponding to the historical charging data.

[0077] In some optional embodiments, the historical charging data of the target vehicle can delimit a high-frequency charging interval as a feature extraction interval based on the type of power battery, the characteristics of the target vehicle, and the user's usage habits to extract historical charging segment data. Among them, the user's usage habits here are not limited to charging, driving, etc. The high-frequency charging interval can be determined by the voltage of the target vehicle from the moment when the voltage reaches during fast charging to the moment when charging is completed.

[0078] Specifically, the maximum charging voltage and the minimum charging voltage of the charging feature extraction interval can be obtained; taking the moment corresponding to the minimum charging voltage as the feature extraction start moment and the moment corresponding to the maximum charging voltage as the feature extraction end moment, the historical charging data is extracted for charging segment data to obtain the historical charging data segment.

[0079] Taking a certain vehicle model as an example, the type of power battery carried by this vehicle model is a ternary lithium battery, and the relationship between its open-circuit voltage OCV and the state of charge SOC of the battery, that is, the OCV-SOC correspondence is as shown in the appendix Figure 2 shown. Statistics of the user's charging behavior, as shown in the appendix Figure 3As shown, a large number of charging behaviors of the user span the state of charge (SOC) range of the battery [60%, 98%], and the corresponding battery voltage range is [3800 mV, 4250 mV]. Therefore, the feature extraction range Vs = [3800 mV, 4250 mV] is defined to ensure that there is sufficient charging data available for feature extraction and then for predicting the battery state. At the same time, during the application stage, when the user's charging behavior also spans this range, feature extraction can be performed to achieve the prediction of the battery health state.

[0080] As Figure 4 shown, for a single charging behavior of the target vehicle, a segment of historical charging data can be obtained. Among them, the time t1 is the moment when the highest single-cell voltage first reaches 3800 mV, and the time tn is the moment when the highest single-cell voltage first reaches 4250 mV. Feature extraction is performed starting from t1 and ending at tn to obtain the historical charging segment data D = [D t , D t2 , …, D tn , where D i is the charging data at the time ti, which can include charging voltage, charging current, and charging temperature. For different charging data samples, the length of the historical charging segment data is different.

[0081] Step S103: Perform feature extraction on the historical charging segment data to obtain the attribute features of the battery in the target vehicle.

[0082] Among them, when performing feature extraction on the historical charging segment data of the target vehicle, the voltage feature of the battery in the target vehicle can be extracted only, and this voltage feature can be used as the attribute feature of the battery; the charge increment feature of the battery in the target vehicle can be extracted only, and this charge increment feature can be used as the attribute feature of the battery; the temperature feature of the battery in the target vehicle can be extracted only, and this temperature feature can be used as the attribute feature of the battery; the voltage feature, the charge increment feature, and the temperature feature can be combined pairwise as the attribute feature of the battery; or the combination of the voltage feature, the charge increment feature, and the temperature feature can be used as the attribute feature of the battery. When the combination of the voltage feature, the charge increment feature, and the temperature feature is used as the attribute feature of the battery, the prediction result of the battery state is more accurate.

[0083] In some alternative embodiments, since the state of the battery pack is affected by the "barrel effect" among battery cells, and the amount of electricity that the battery pack can store and release, as determined by the battery management system, is also affected by the voltage states of individual cells, voltage characteristics can be used as attribute characteristics to extract effective battery cell information therefrom, and then the battery state can be predicted. Therefore, when extracting the attribute characteristics of the battery in the target vehicle from the historical charging segment data, the first voltage sampling of the battery cells can be performed on the historical charging segment data to obtain the individual voltage sequences of the battery cells in the battery; the individual voltage sequences can be concatenated to obtain the voltage characteristics of the battery, that is, the voltage characteristics are used as the attribute characteristics.

[0084] Taking the above vehicle model as an example, the power battery system of the target vehicle includes 96 battery cells. In the extracted historical charging segment data D, the first voltage sampling of each battery cell is performed on the historical charging segment data. For example, uniform sampling in the time dimension is performed on the charging voltage points, and the number of sampling points is 96. The number of sampling points can be adjusted according to actual needs, so as to obtain the individual voltage sequences of 96 battery cells, and the individual voltage sequences are concatenated to obtain the matrix V of the voltage characteristics. c ,

[0085]

[0086] where V i,j represents the i-th voltage sampling value of the j-th individual battery. As Figure 5 shown, the voltages of individual cells gradually increase during charging, but the voltage increase rates are inconsistent, and there are voltage differences among battery cells. The matrix V of the voltage characteristics c can reflect the inconsistency during the battery charging process and has a representative effect on the overall state of the battery pack.

[0087] The battery health state prediction method provided by the embodiments of the present invention obtains the individual voltage sequences of the battery cells in the battery by performing the first voltage sampling of the battery cells on the historical charging segment data, so as to obtain the voltage characteristics of the battery, providing a basis for the determination of the fusion characteristics and the determination of the battery state (health state).

[0088] In some alternative embodiments, during the charging process, for battery cells in a relatively poor state, their voltages rise more significantly when the same amount of charge is input. At the same time, to protect the battery cells in a relatively poor state from overcharging, the battery management system often controls the charging behavior based on the highest cell voltage. As the health state of the battery deteriorates, the charge increment curve based on the highest cell voltage gradually shifts downward. Therefore, when extracting the attribute features of the battery in the target vehicle from the historical charging segment data, the second voltage sampling of the battery cells can be performed on the historical charging segment data to obtain the highest cell voltage of all battery cells at each sampling moment, forming the highest cell voltage sequence. Based on the highest cell voltage sequence, the corresponding highest cell current sequence is obtained to obtain the charge increment at each sampling moment. Interpolation is performed based on the charge increments at each sampling moment to obtain the charge increments corresponding to each voltage point in the historical charging segment data, forming the charge increment sequence, that is, taking the charge increment sequence as the attribute feature.

[0089] Taking the above vehicle model as an example, from the extracted historical charging segment data D, the highest cell voltage data at each sampling moment is extracted to construct the highest cell voltage sequence V max =[V t1 ,V t2 ,…,V ti ,…,V tn ; Based on the highest cell voltage sequence V max =[V t1 ,V t2 ,…,V ti ,…,V tn , the corresponding highest cell current sequence is obtained, and the charge increment Q t , Q t =[Q t1 ,Q t2 ,…,Q ti ,…,Q tn of the historical charging segment data D at each sampling moment is calculated by the ampere-hour integration method. Among them, the calculation method of Q ti is as follows:

[0090]

[0091] Among them, I is the charging current. Since the charging duration and sequence length are different for each charging, length normalization is required. The Q t sequence is interpolated to the individual voltage points V t =[3801mV, 3802mV, …, 4250mV] in the historical charging segment data to obtain the charge increment corresponding to each voltage point; and then the charge increment sequence Q, Q = [Q1, Q2, …, Q 450 is formed. The interpolation process is as follows:

[0092]

[0093] The battery state of health prediction method provided by the embodiment of the present invention performs a second voltage sampling on each battery cell for the historical charging segment data, obtains the highest cell voltage of all battery cells at each sampling moment, forms a highest cell voltage sequence, and obtains the corresponding highest cell current sequence; obtains the corresponding highest cell current sequence based on the highest cell voltage sequence to calculate the power increment at each sampling moment; performs interpolation based on the power increment at each sampling moment to obtain the power increment corresponding to each voltage point in the historical charging segment data, forms a power increment sequence, and provides a basis for the determination of the fusion feature and the determination of the battery state of health.

[0094] In some optional embodiments, the state of health of the battery is greatly affected by temperature, and the temperature sequence can effectively help predict the current state of health of the battery. Therefore, when extracting the attribute features of the battery in the target vehicle by performing feature extraction on the historical charging segment data, the temperature of the battery at each sampling moment can be obtained to obtain a temperature sequence, that is, the temperature sequence is used as an attribute feature.

[0095] Taking the above vehicle model as an example, there are multiple temperature sensors in the battery system. When extracting the historical charging segment data D, calculate the average temperature of all sensors and perform uniform sampling in time. The number of sampling points corresponding to each sampling moment is 450, and a temperature sequence T is constructed, T = [T1, T2, …, T 450 .

[0096] The battery state of health prediction method provided by the embodiment of the present invention obtains a temperature sequence by obtaining the temperature of the battery at each sampling moment, and provides a basis for the determination of the fusion feature and the determination of the battery state of health.

[0097] Step S104, perform feature analysis based on the historical driving data and the attribute features to obtain a fusion feature.

[0098] Among them, perform feature analysis based on the historical driving data and the attribute features to obtain the cumulative driving mileage and service life of the historical driving data, and at the same time perform feature fusion on the attribute features to obtain a fusion feature.

[0099] In some optional embodiments, when performing feature analysis based on the historical driving data and the attribute features to obtain the fusion feature of the battery, statistical analysis can be performed on the voltage feature, the power increment sequence, and the temperature sequence respectively to obtain the corresponding statistical analysis features; based on the fusion of the statistical analysis features and the historical driving data, a fusion feature is obtained.

[0100] Taking the above vehicle model as an example, a matrix V of the voltage feature can be cPerform statistical analysis to extract the maximum single-cell voltage range as feature point 1 from the matrix V of voltage characteristics; perform statistical analysis on the sequence of power increments to extract the mean, range, and standard deviation as feature points 2-4 from the sequence of power increments Q, and perform statistical analysis on the temperature sequence to extract the average temperature, maximum temperature, and minimum temperature as feature points 5-7; extract the cumulative driving mileage and service life from the historical driving data as feature points 8-9. Obtain the fusion feature based on feature points 1-9. Among them, when performing statistical analysis on the voltage characteristics, the sequence of power increments, and the temperature sequence, the selected feature points can be increased or decreased based on actual requirements. c The battery health state prediction method provided by the embodiment of the present invention obtains corresponding statistical analysis features by respectively performing statistical analysis on the voltage characteristics, the sequence of power increments, and the temperature sequence, fuses the statistical analysis features with the historical driving data to obtain the fusion feature, and predicts the state of the battery based on the fusion feature, and improves the accuracy of predicting the battery health state.

[0101] Step S105, predict the state of the battery based on the fusion feature to obtain the state prediction result of the battery.

[0102] When predicting the state of the battery based on the fusion feature, it is based on the construction of a machine learning model. This machine learning model is not limited to the machine regression learning model. When using the machine regression learning model, it can be not limited to the Gaussian process regression model, support vector machine model, relevance vector machine model, multiple linear regression model, multi-order polynomial model, random forest model, etc.

[0103] In some optional embodiments, when predicting the state of the battery based on the fusion feature to obtain the state prediction result of the battery, the fusion feature can be input into the first state prediction model to obtain the first state prediction result; determine the state prediction result of the battery based on the first state prediction result. That is, the first state prediction result can be directly used as the state prediction result of the battery; or on the basis of the first state prediction result, in combination with the first state prediction results of other prediction models, the first state prediction result and the second state prediction result are fused to obtain the state prediction result of the battery.

[0104]

[0105] ​Taking the direct use of the first state prediction result as the battery state prediction result as an example, if the first state prediction model is a Gaussian process regression model, for instance, points features 1 to 9 are used as inputs, and the battery state is used as the output. Machine learning is performed to obtain the first state prediction result, and the first state prediction result is used as the battery state prediction result, improving the accuracy of predicting the battery health state. Of course, the first state prediction model is not limited to the above-mentioned Gaussian process regression model, and it can also be a support vector machine model, a relevance vector machine model, a multiple linear regression model, a multi-order polynomial model, a random forest model, etc., which are not restricted here.

[0106] The battery health state prediction method provided in this embodiment obtains the historical charging data of the target vehicle and the corresponding historical driving data, analyzes the historical charging data, and further obtains the historical charging segment data that can balance the feature effectiveness and availability; extracts the attribute features of the battery in the target vehicle by performing feature extraction on the historical charging segment data to improve the accuracy of predicting the battery health state; obtains the fusion features by performing feature analysis based on the historical driving data and the attribute features, thereby further improving the accuracy of predicting the battery health state; predicts the state of the battery based on the fusion features to obtain the battery state prediction result, so as to facilitate timely reminding the user to perform battery maintenance and replacement based on the state prediction result.

[0107] In this embodiment, a battery health state prediction method is provided. Figure 5 It is a flowchart of the battery health state prediction method according to an embodiment of the present invention. As Figure 6 shown, this process includes the following steps:

[0108] Step S601, obtain the historical charging data of the target vehicle and the corresponding historical driving data.

[0109] For details, please refer to Figure 1 step S101 of the embodiment shown, which will not be elaborated here.

[0110] Step S602, analyze the historical charging data to obtain the historical charging segment data.

[0111] For details, please refer to Figure 1 step S102 of the embodiment shown, which will not be elaborated here.

[0112] Step S603, perform feature extraction on the historical charging segment data to obtain the attribute features of the battery in the target vehicle.

[0113] For details, please refer to Figure 1 step S103 of the embodiment shown, which will not be elaborated here.

[0114] Step S604: Perform feature analysis based on historical driving data and attribute features to obtain fused features.

[0115] For details, please refer to Figure 1 step S104 of the illustrated embodiment, which will not be elaborated here.

[0116] Step S605: Predict the state of the battery based on the fused features to obtain the battery state prediction result.

[0117] Among them, the attribute features include the voltage feature of the battery and the time feature sequence, and the time feature sequence includes the power increment sequence and the temperature sequence.

[0118] Specifically, the above step S605 includes:

[0119] Step S6051: Obtain the fourth feature in the first state prediction result.

[0120] Among them, the fourth feature includes the probability average distribution mean and the standard deviation. That is, the probability average distribution mean and the standard deviation are partial features representing the battery health state obtained by machine learning (such as machine regression learning) of the fused features.

[0121] Step S6052: Input the voltage feature, the time feature sequence, and the fused features into the feature extraction unit corresponding to the second state prediction model to obtain the corresponding first feature, second feature, and third feature.

[0122] Among them, the voltage feature, the time feature sequence, and the fused features all have their respective feature extraction units in the second state prediction model. That is, the feature extraction unit corresponding to the voltage feature is a feature extraction unit adapted to the voltage feature, the feature extraction unit corresponding to the time sequence time feature sequence is a feature extraction unit adapted to the time sequence time feature sequence, and the feature extraction unit corresponding to the fused features is a feature extraction unit adapted to the fused features.

[0123] Exemplarily, such as Figure 7As shown, the second state prediction model is constructed based on a deep learning neural network. The feature extraction unit in the second state prediction model includes a Convolutional Neural Networks (CNN) feature extraction unit, a Long Short-Term Memory (LSTM) feature extraction unit, and a Fully Convolutional Networks (FCN) feature extraction unit. The Convolutional Neural Networks (CNN) feature extraction unit is used to extract features of voltage features to obtain corresponding first features; the Long Short-Term Memory (LSTM) feature extraction unit is used to extract features of the time series time feature sequence to obtain corresponding second features; the Fully Convolutional Networks (FCN) feature extraction unit is used to extract features of fused features to obtain corresponding third features. Among them, in other examples, the feature extraction unit included in the second state prediction model can be adjusted based on the attributes of voltage features, time feature sequences, and fused features.

[0124] Specifically, the matrix V of voltage features c is input into the Convolutional Neural Networks (CNN) feature extraction unit for feature extraction to obtain corresponding first features; the power increment sequence Q and the temperature sequence T are input into the Long Short-Term Memory (LSTM) feature extraction unit for feature extraction to obtain corresponding second features; the feature points 1 to 9 are input into the Fully Convolutional Networks (FCN) feature extraction unit for feature extraction to obtain third features.

[0125] Step S6053, input the first feature, the second feature, the third feature, and the fourth feature into the prediction unit of the second state prediction model to obtain the second state prediction result.

[0126] Specifically, the first feature, the second feature, the third feature, and the fourth feature can be concatenated first, and then the concatenated features are input into the prediction unit of the second state prediction model for linear regression (i.e., linear summation) to obtain the second state prediction result. It is also possible to input the first feature, the second feature, the third feature, and the fourth feature into the prediction unit of the second state prediction model for feature concatenation and then perform sequential linear regression calculations to obtain the second state prediction result.

[0127] Alternatively, the prediction unit can also first obtain the weights corresponding to the convolutional neural network (CNN) feature extraction unit, the long short-term memory neural network (LSTM) feature extraction unit, and the fully connected layer network (FCN) feature extraction unit. Based on the weights corresponding to the convolutional neural network (CNN) feature extraction unit, the long short-term memory neural network (LSTM) feature extraction unit, and the fully connected layer network (FCN) feature extraction unit, the first feature, the second feature, and the third feature are first fused and then input into the prediction unit (such as the fully connected layer regression calculation unit) of the second state prediction model for prediction to obtain the second state prediction result.

[0128] In some alternative embodiments, the voltage feature, the time feature sequence, and the fusion feature can be input into the feature extraction unit corresponding to the second state prediction model to obtain the corresponding first feature, second feature, and third feature. The first feature, the second feature, and the third feature are input into the prediction unit of the second state prediction model, and the obtained second state prediction result can also be directly used as the battery state prediction result.

[0129] Specifically, the first feature, the second feature, and the third feature can be concatenated first, and then the concatenated feature is input into the feature extraction unit corresponding to the second state prediction model for linear regression (i.e., linear summation) to obtain the second prediction result. The first feature, the second feature, and the third feature can also be input into the prediction unit of the second state prediction model for feature concatenation and then sequential linear regression calculation to obtain the second state prediction result.

[0130] Alternatively, the prediction unit can also first obtain the weights corresponding to the convolutional neural network (CNN) feature extraction unit, the long short-term memory neural network (LSTM) feature extraction unit, and the fully connected layer network (FCN) feature extraction unit. Based on the weights corresponding to the convolutional neural network (CNN) feature extraction unit, the long short-term memory neural network (LSTM) feature extraction unit, and the fully connected layer network (FCN) feature extraction unit, the first feature, the second feature, and the third feature are first fused and then input into the prediction unit (such as the fully connected layer regression calculation unit) of the second state prediction model for prediction to obtain the second state prediction result.

[0131] Step S6054, determining the battery state prediction result based on the second state prediction result.

[0132] By collaborative learning based on the first state prediction model and the second state prediction model, the second state prediction result is obtained, and the battery health state prediction result is determined based on the second state prediction result, further improving the accuracy and efficiency of the battery health state prediction.

[0133] In some alternative embodiments, the first state prediction result and the second state prediction result may also be fused to obtain the state prediction result of the battery.

[0134] Specifically, the first weight corresponding to the first state prediction model and the second weight corresponding to the second state prediction model may be obtained first; based on the first weight and the second weight, the corresponding weighted fusion of the first state prediction result and the second state prediction result is performed to obtain the state prediction result of the battery. By using the first weight corresponding to the first state prediction model and the second weight corresponding to the second state prediction model, the reliability of the fusion of the first state prediction result and the second state prediction result is improved.

[0135] In some alternative embodiments, both the first state prediction result and the second state prediction result include probability mean distribution and standard deviation. The weight corresponding to the first state prediction model is used as the first weight, and the weight corresponding to the second state prediction model is used as the second weight. Based on the first weight and the second weight, the corresponding weighted fusion of the first state prediction result (weights of probability mean distribution and standard deviation) and the second state prediction result (weights of probability mean distribution and standard deviation) is performed to obtain the state prediction result of the battery.

[0136] The battery health state prediction method provided by the embodiments of the present invention further improves the dimension and accuracy of the extraction of battery state-related features by obtaining the fourth feature of the first state prediction result and inputting the voltage feature, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model to obtain the corresponding first feature, second feature, and third feature; and further improves the accuracy of the battery health state prediction by performing state prediction on the concatenation result of the first feature, the second feature, the third feature, and the fourth feature through the prediction unit of the second state prediction model to obtain the second state prediction result.

[0137] In this embodiment, a battery health state prediction device is further provided. This device is used to implement the above embodiments and preferred implementation manners, and those that have been described will not be repeated. As used below, the term "module" may be a combination of software and / or hardware that can implement a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0138] This embodiment provides a battery health state prediction device, as Figure 8 shown, including:

[0139] A data acquisition module 801, configured to acquire historical charging data of a target vehicle and historical driving data corresponding to the historical charging data.

[0140] The data analysis module 802 is used to analyze historical charging data to obtain historical charging segment data;

[0141] The feature extraction module 803 is used to extract features from the historical charging segment data to obtain the attribute features of the battery in the target vehicle.

[0142] The feature analysis module 804 is used to perform feature analysis based on historical driving data and attribute features to obtain fused features.

[0143] The state prediction module 805 is used to predict the state of the battery based on the fused features to obtain the battery state prediction result.

[0144] In some alternative embodiments, the feature extraction module 803 includes:

[0145] The first sampling unit is used to perform the first voltage sampling of battery cells on the historical charging segment data to obtain the cell voltage sequence of each battery cell in the battery;

[0146] The first splicing unit is used to splice the cell voltage sequence to obtain the voltage feature of the battery, and the attribute features include the voltage feature.

[0147] In some alternative embodiments, the feature extraction module 803 includes:

[0148] The second sampling unit is used to perform the second voltage sampling of battery cells on the historical charging segment data to obtain the highest cell voltage of all battery cells at each sampling moment, forming the highest cell voltage sequence;

[0149] The sequence acquisition unit is used to obtain the corresponding highest cell current sequence based on the highest cell voltage sequence to obtain the power increment at each sampling moment;

[0150] The power interpolation unit is used to perform interpolation based on the power increment at each sampling moment to obtain the power increment corresponding to each voltage point in the historical charging segment data, forming the power increment sequence, and the attribute features include the power increment sequence.

[0151] In some alternative embodiments, the attribute features include the voltage feature of the battery, the power increment sequence, and the temperature sequence. The feature extraction module 803 includes:

[0152] The temperature acquisition unit acquires the temperature of the battery at each sampling moment to obtain the temperature sequence, and the attribute features include the temperature sequence.

[0153] In some alternative embodiments, the feature analysis module 804 includes:

[0154] A statistical analysis unit for performing statistical analysis on the voltage characteristics, the power increment sequence, and the temperature sequence respectively to obtain corresponding statistical analysis features;

[0155] A data fusion unit for obtaining a fusion feature based on the fusion of the statistical analysis features and the historical driving data.

[0156] In some alternative embodiments, the state prediction module 805 includes:

[0157] A state prediction unit for inputting the fusion feature into a first state prediction model to obtain a first state prediction result;

[0158] A result determination unit for determining the state prediction result of the battery based on the first state prediction result.

[0159] In some alternative embodiments, the attribute features include the voltage characteristics of the battery and the time feature sequence, and the time feature sequence includes the power increment sequence and the temperature sequence. The result determination unit includes:

[0160] A feature acquisition subunit for acquiring a fourth feature in the first state prediction result;

[0161] A feature input subunit for inputting the voltage characteristics, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model to obtain corresponding first, second, and third features;

[0162] A state prediction subunit for inputting the first, second, third, and fourth features into the prediction unit of the second state prediction model to obtain a second state prediction result;

[0163] A state determination subunit for determining the state prediction result of the battery based on the second state prediction result.

[0164] In some alternative embodiments, the attribute features include the voltage characteristics of the battery and the time feature sequence, and the time feature sequence includes the power increment sequence and the temperature sequence. The result determination unit includes:

[0165] A feature input subunit for inputting the voltage characteristics, the time feature sequence, and the fusion feature into the feature extraction unit corresponding to the second state prediction model to obtain corresponding first, second, and third features;

[0166] A state prediction subunit for inputting the first, second, and third features into the prediction unit of the second state prediction model to obtain a second state prediction result;

[0167] A feature fusion subunit for fusing the first state prediction result and the second state prediction result to obtain the state prediction result of the battery.

[0168] A result determination subunit, configured to determine a state prediction result of the battery based on the second state prediction result.

[0169] In some optional embodiments, the first state prediction model is constructed based on a machine learning model, and the second state prediction model is constructed based on a deep learning neural network.

[0170] The battery health state prediction device provided in this embodiment obtains historical charging data of a target vehicle and historical driving data corresponding to the historical charging data, analyzes the historical charging data, and further obtains historical charging segment data that can balance feature effectiveness and availability; extracts features from the historical charging segment data to obtain attribute features of the battery in the target vehicle, so as to improve the accuracy of predicting the battery health state; performs feature analysis based on the historical driving data and the attribute features to obtain fused features, thereby further improving the accuracy of predicting the battery health state; predicts the state of the battery based on the fused features to obtain a state prediction result of the battery, so as to facilitate timely reminding the user to perform battery maintenance and replacement based on the state prediction result.

[0171] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding embodiments above, and will not be repeated here.

[0172] The battery health state prediction device in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0173] An embodiment of the present invention further provides a computer device having the above-mentioned Figure 8 battery health state prediction device shown.

[0174] Please refer to Figure 9 , Figure 9 which is a schematic structural diagram of a computer device provided by an optional embodiment of the present invention. As shown in Figure 9As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Each component communicates with each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory 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 alternative embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a set of blade servers, or a multi-processor system). Figure 9 Taking one processor 10 as an example in

[0175] The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above programmable logic device can be a complex programmable logic device, a field programmable gate array, a generic array logic, or any combination thereof.

[0176] Among them, 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 embodiments.

[0177] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely set relative to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0178] The memory 20 can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memories.

[0179] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

[0180] Embodiments of the present invention also provide a computer-readable storage medium. The method according to the embodiments of the present invention can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, 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 memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, 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, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0181] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for predicting the state of health of a battery, characterized in that, The method includes: Obtaining historical charging data of a target vehicle and historical driving data corresponding to the historical charging data; Analyzing the historical charging data to obtain historical charging segment data; Extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle; Performing feature analysis based on the historical driving data and the attribute features to obtain fused features; Predicting the state of the battery based on the fused features to obtain a state prediction result of the battery.

2. The method according to claim 1, wherein The extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle includes: Performing first voltage sampling on battery cells of the historical charging segment data to obtain a cell voltage sequence of each battery cell in the battery; Concatenating the cell voltage sequences to obtain a voltage feature of the battery, and the attribute features include the voltage feature.

3. The method according to claim 2, wherein The extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle includes: Performing second voltage sampling on battery cells of the historical charging segment data to obtain the highest cell voltage of all the battery cells at each sampling moment, and forming a highest cell voltage sequence; Obtaining a corresponding highest cell current sequence based on the highest cell voltage sequence to obtain the power increment at each sampling moment; Performing interpolation based on the power increments at each sampling moment to obtain the power increment corresponding to each voltage point in the historical charging segment data, and forming a power increment sequence, and the attribute features include the power increment sequence.

4. The method according to claim 3, wherein The extracting features from the historical charging segment data to obtain attribute features of the battery in the target vehicle includes: Obtaining the temperature of the battery at each sampling moment to obtain a temperature sequence, and the attribute features include the temperature sequence.

5. The method according to claim 1, wherein The attribute features include the voltage feature of the battery, the power increment sequence, and the temperature sequence. The performing feature analysis based on the historical driving data and the attribute features to obtain the fused features of the battery includes: Performing statistical analysis on the voltage feature, the power increment sequence, and the temperature sequence respectively to obtain corresponding statistical analysis features; Obtaining the fused features based on the fusion of the statistical analysis features and the historical driving data.

6. The method according to any one of claims 1 to 5, characterized in that The predicting the state of the battery based on the fused features to obtain a state prediction result of the battery includes: Inputting the fused features into a first state prediction model to obtain a first state prediction result; Determining the state prediction result of the battery based on the first state prediction result.

7. The method according to claim 6, characterized in that, The attribute features include the voltage feature of the battery and a time feature sequence, and the time feature sequence includes the power increment sequence and the temperature sequence. The determining the state prediction result of the battery based on the first state prediction result includes: Obtaining a fourth feature in the first state prediction result; Inputting the voltage feature, the time feature sequence, and the fused features into a feature extraction unit corresponding to a second state prediction model to obtain corresponding first, second, and third features; Input the first feature, the second feature, the third feature, and the fourth feature into the prediction unit of the second state prediction model to obtain a second state prediction result; Determine the state prediction result of the battery based on the second state prediction result.

8. The method according to claim 7, characterized in that The first state prediction model is constructed based on a machine learning model, and the second state prediction model is constructed based on a deep learning neural network.

9. The method according to any one of claims 1 to 5, characterized in that The attribute features include the voltage feature of the battery and a time feature sequence. The time feature sequence includes a power increment sequence and a temperature sequence. Predicting the state of the battery based on the fusion features to obtain the state prediction result of the battery includes: Input the voltage feature, the time feature sequence, and the fusion features into the corresponding feature extraction unit of the second state prediction model to obtain corresponding first, second, and third features; Input the first feature, the second feature, and the third feature into the prediction unit of the second state prediction model to obtain a second state prediction result; Determine the state prediction result of the battery based on the second state prediction result.

10. A battery state of health prediction device, characterized in that, The device includes: A data acquisition module for acquiring historical charging data of a target vehicle and historical driving data corresponding to the historical charging data; A data analysis module for analyzing the historical charging data to obtain historical charging segment data; A feature extraction module for extracting features from the historical charging segment data to obtain the attribute features of the battery in the target vehicle; A feature analysis module for performing feature analysis based on the historical driving data and the attribute features to obtain fusion features; A state prediction module for predicting the state of the battery based on the fusion features to obtain the state prediction result of the battery.

11. A computer device, characterized in that, Includes: A memory and a processor. The memory and the processor are communicatively connected to each other. The memory stores computer instructions. The processor executes the computer instructions to execute the battery health state prediction method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer-readable storage medium. The computer instructions are used to cause a computer to execute the battery health state prediction method according to any one of claims 1 to 9.

Images