Battery life prediction method, device, server, and storage medium

Abstract

This application relates to the field of electric vehicle technology, and particularly to a battery life prediction method, device, server, and storage medium. The method includes: acquiring vehicle data that has reached a battery life degradation node; identifying basic vehicle attribute information and internal / external battery usage information from the vehicle data; establishing a first indicator dataset based on the basic attribute information and a second indicator dataset based on the internal / external battery usage information; assigning weights to the indicators in the second indicator dataset according to time stages, and training a first prediction model by combining the battery cycle count corresponding to the battery life degradation node; after training, associating the first indicator dataset with the first prediction model, and using the first prediction model to predict the vehicle's battery life. This solves the problem in related technologies where insufficient consideration of internal / external battery usage information leads to poor prediction accuracy when predicting battery life.

Description

Technical Field

[0001] This application relates to the field of electric vehicle technology, and in particular to a battery life prediction method, device, server, and storage medium. Background Technology

[0002] With the continuous development of electric vehicle technology, the importance of batteries as their power source is self-evident. However, as batteries are used, their lifespan will continue to shorten, making battery life prediction increasingly important. Currently, the following technologies exist in the field of battery life prediction:

[0003] (1) Lithium-ion battery life prediction system based on random forest regression (Publication No.: CN112560287A): The battery life is defined as the number of cycles corresponding to SOH (State of Health) <= 80%. Data collected inside the battery, such as voltage drop rate, initial capacity, resistance and temperature, are used as influencing factors. Factors obtained through correlation analysis are fed into the random forest model for training to obtain the battery life prediction model. This related technology can use machine learning models to fit the relationship between battery electrochemical factors and life results. However, it does not consider how battery life as a future degradation value is represented in features and models in actual business.

[0004] (2) Lithium battery life prediction method based on time series features (Publication No.: CN114861527A): By mining the charging and discharging time series features and establishing a vector group, a GRU (Gated Recurrent Unit) neural network training model is established to obtain the prediction effect. This related technology uses a model with time series fitting ability to handle the nonlinear trend of battery capacity in the life cycle, but does not take into account the influence of external factors. Summary of the Invention

[0005] This application provides a battery life prediction method, apparatus, server, and storage medium to solve the problems in related technologies where the prediction of battery life does not fully consider internal and external information during battery use, resulting in poor accuracy of the prediction results.

[0006] The first aspect of this application provides a battery life prediction method, comprising the following steps: acquiring vehicle data that has reached a battery life degradation node; identifying basic vehicle attribute information and internal and external battery usage information in the vehicle data; establishing a first indicator dataset based on the basic attribute information; establishing a second indicator dataset based on the internal and external battery usage information; assigning weights to indicators in the second indicator dataset according to time stages; training a first prediction model in conjunction with the battery cycle count corresponding to the battery life degradation node; associating the first indicator dataset with the first prediction model after training; matching the first prediction model of the vehicle based on the basic attribute information; inputting the internal and external battery usage information of the vehicle into the first prediction model; and outputting the first prediction cycle count corresponding to the battery life degradation node of the vehicle.

[0007] Based on the above technical means, the embodiments of this application can establish first and second indicator datasets based on the basic attribute information of the vehicle and the internal and external information of the battery usage process, and train the first prediction model by combining the number of battery cycles corresponding to the battery life decay node. By constructing a rolling time series indicator system to cover multiple factors inside and outside the battery, the model can be guaranteed to absorb as many dimensional information as possible. In addition, the indicators in the second indicator dataset are weighted according to the time stage, which can meet the nonlinear trend of the battery mechanism, thereby effectively improving the accuracy of battery life prediction.

[0008] Optionally, in one embodiment of this application, the internal and external information of battery usage includes one or more of battery stability information, user driving information, and external operating condition information, and the second indicator dataset includes one or more of the indicator datasets of battery stability information, user driving information, and external operating condition information.

[0009] Optionally, in one embodiment of this application, the step of establishing a second indicator dataset based on the internal and external information of the battery usage process includes: obtaining the battery usage time; dividing the usage time into multiple time stages, and calculating the differences of all indicators in the internal and external information of the battery usage process in the order of the time stages; and generating the second indicator dataset based on the differences of all indicators in all time stages.

[0010] Based on the above technical means, the embodiments of this application can calculate the differences of all indicators in the internal and external information of battery use at different time stages based on the battery usage time, and generate a second indicator dataset based on the differences, which can effectively improve the data accuracy of the second indicator dataset.

[0011] Optionally, in one embodiment of this application, the weight values ​​of the indicators in the second indicator dataset are set from small to large according to the chronological order of time stages.

[0012] Based on the above technical means, the embodiments of this application can set the weight values ​​of the indicators in the second indicator dataset in ascending order of time stages, which can effectively improve the practicality and accuracy of the dataset.

[0013] Optionally, in one embodiment of this application, after inputting the internal and external information of the vehicle's battery usage process into the first prediction model and outputting the first predicted cycle number corresponding to the battery life decay node of the vehicle, the method further includes: matching the vehicle's second prediction model according to the basic attribute information; inputting the vehicle's battery health status (SOH) and actual cycle number into the second prediction model and outputting the second predicted cycle number corresponding to the battery life decay node of the vehicle; and using the second predicted cycle number to correct the first predicted cycle number to obtain the third predicted cycle number corresponding to the battery life decay node of the vehicle.

[0014] Based on the above technical means, the embodiments of this application can obtain the second predicted number of cycles corresponding to the battery life decay node of the vehicle according to the second prediction model after obtaining the first predicted number of cycles, and correct the first predicted number of cycles according to the second predicted number of cycles to obtain the third predicted number of cycles, which can effectively improve the accuracy of the predicted number of cycles.

[0015] Optionally, in one embodiment of this application, the step of correcting the first predicted cycle number using the second predicted cycle number to obtain the third predicted cycle number corresponding to the battery life degradation node of the vehicle includes: identifying the rated cycle number of the battery in the basic attribute information; and calculating the third predicted cycle number based on the rated cycle number, the first predicted cycle number, and the second predicted cycle number.

[0016] Based on the above technical means, the embodiments of this application can calculate the third predicted cycle number based on the battery's rated cycle number, the first predicted cycle number, and the second predicted cycle number in the basic attribute information, which can effectively improve the accuracy of the predicted cycle number.

[0017] Optionally, in one embodiment of this application, the construction process of the second prediction model includes: acquiring training data, wherein the training data includes SOH and the number of iterations; constructing a nonlinear relationship curve between SOH and the number of iterations based on the training data, and clustering the nonlinear relationship curve based on the basic attribute information; establishing a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, and associating the first index dataset with the second prediction model.

[0018] Based on the above technical means, the embodiments of this application can construct nonlinear relationship curves based on training data and perform clustering, and then establish a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, which can effectively realize the real-time updating of the prediction cycle number of vehicles.

[0019] A second aspect of this application provides a battery life prediction method, comprising the following steps: acquiring basic attribute information of a vehicle and internal and external information of battery usage; matching a first prediction model of the vehicle according to the basic attribute information; inputting the internal and external information of battery usage of the vehicle into the first prediction model; and outputting a first prediction cycle number corresponding to the battery life decay node of the vehicle, wherein the first prediction model is trained based on vehicle data that has reached the battery life decay node.

[0020] Based on the above technical means, the embodiments of this application can match the vehicle's first prediction model based on the vehicle's basic attribute information, and obtain the first prediction cycle number corresponding to the battery life decay node of the vehicle by inputting internal and external information of the battery usage process, which can effectively improve the accuracy of the prediction cycle number.

[0021] A third aspect of this application provides a battery life prediction device, comprising: a first acquisition module for acquiring vehicle data that has reached a battery life degradation node; an establishment module for identifying basic attribute information of the vehicle and internal and external information of battery usage in the vehicle data, establishing a first index dataset based on the basic attribute information, and establishing a second index dataset based on the internal and external information of battery usage; a training module for assigning weights to the indicators in the second index dataset according to time stages, and training a first prediction model in conjunction with the number of battery cycles corresponding to the battery life degradation node, and associating the first index dataset with the first prediction model after training; and a matching module for matching the first prediction model of the vehicle based on the basic attribute information, inputting the internal and external information of battery usage of the vehicle into the first prediction model, and outputting the first prediction cycle number corresponding to the battery life degradation node of the vehicle.

[0022] Optionally, in one embodiment of this application, the internal and external information of battery usage includes one or more of battery stability information, user driving information, and external operating condition information, and the second indicator dataset includes one or more of the indicator datasets of battery stability information, user driving information, and external operating condition information.

[0023] Optionally, in one embodiment of this application, the establishment module is further configured to: obtain the battery usage time; divide the usage time into multiple time stages, and calculate the differences of all indicators in the internal and external information of the battery usage process in the order of the time stages; and generate the second indicator dataset based on the differences of all indicators in all time stages.

[0024] Optionally, in one embodiment of this application, the weight values ​​of the indicators in the second indicator dataset are set from small to large according to the chronological order of time stages.

[0025] Optionally, in one embodiment of this application, the battery life prediction device further includes: a correction module, configured to, after inputting internal and external information of the vehicle's battery usage process into the first prediction model and outputting the first predicted cycle number corresponding to the battery life degradation node of the vehicle, match the second prediction model of the vehicle based on the basic attribute information; input the vehicle's battery health status (SOH) and actual cycle number into the second prediction model and output the second predicted cycle number corresponding to the battery life degradation node of the vehicle; and correct the first predicted cycle number using the second predicted cycle number to obtain the third predicted cycle number corresponding to the battery life degradation node of the vehicle.

[0026] Optionally, in one embodiment of this application, the correction module is further configured to: identify the rated cycle number of the battery in the basic attribute information; and calculate the third predicted cycle number based on the rated cycle number, the first predicted cycle number, and the second predicted cycle number.

[0027] Optionally, in one embodiment of this application, the correction module may further be used to: acquire training data, wherein the training data includes SOH and the number of iterations; construct a nonlinear relationship curve between SOH and the number of iterations based on the training data, and cluster the nonlinear relationship curve based on the basic attribute information; establish a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, and associate the first index dataset with the second prediction model.

[0028] A fourth aspect of this application provides a battery life prediction device, comprising: a second acquisition module for acquiring basic attribute information of a vehicle and internal and external information of battery usage; and a prediction module for matching a first prediction model of the vehicle based on the basic attribute information, inputting the internal and external information of battery usage of the vehicle into the first prediction model, and outputting a first prediction cycle number corresponding to the battery life decay node of the vehicle, wherein the first prediction model is trained based on vehicle data that has reached the battery life decay node.

[0029] A fifth aspect of this application provides a server, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the battery life prediction method as described in the above embodiments.

[0030] A sixth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the battery life prediction method as described in the above embodiments.

[0031] Therefore, this application has at least the following beneficial effects:

[0032] 1. The embodiments of this application can establish first and second indicator datasets based on the basic attribute information of the vehicle and the internal and external information of the battery usage process, and train the first prediction model by combining the number of battery cycles corresponding to the battery life decay node. By constructing a rolling time series indicator system to cover multiple factors inside and outside the battery, the model can be guaranteed to absorb as many dimensions of information as possible. In addition, the indicators in the second indicator dataset are weighted according to the time stage, which can meet the nonlinear trend of the battery mechanism, thereby effectively improving the accuracy of battery life prediction.

[0033] 2. The embodiments of this application can calculate the differences of all indicators in the internal and external information of the battery usage process at different time stages based on the battery usage time, and generate a second indicator dataset based on the differences, which can effectively improve the data accuracy of the second indicator dataset.

[0034] 3. The embodiments of this application can set the weight values ​​of the indicators in the second indicator dataset in ascending order of time stages, which can effectively improve the practicality and accuracy of the dataset.

[0035] 4. In this embodiment of the application, after obtaining the first predicted cycle number corresponding to the battery life decay node of the vehicle, the second predicted cycle number corresponding to the battery life decay node of the vehicle is obtained according to the second prediction model, and the first predicted cycle number is corrected according to the second predicted cycle number to obtain the third predicted cycle number, which can effectively improve the accuracy of the predicted cycle number.

[0036] 5. The embodiments of this application can calculate the third predicted cycle number based on the battery's rated cycle number, the first predicted cycle number, and the second predicted cycle number in the basic attribute information, which can effectively improve the accuracy of the predicted cycle number.

[0037] 6. The embodiments of this application can construct nonlinear relationship curves based on training data and perform clustering. Then, based on the nonlinear relationship curves of the same clustering type, a corresponding second prediction model can be established, which can effectively realize the real-time updating of the prediction cycle number of vehicles.

[0038] 7. The embodiments of this application can match the vehicle's first prediction model based on the vehicle's basic attribute information, and obtain the first prediction cycle number corresponding to the battery life decay node of the vehicle by inputting internal and external information of the battery usage process, which can effectively improve the accuracy of the prediction cycle number.

[0039] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0040] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0041] Figure 1 This is a flowchart of a battery life prediction method provided according to an embodiment of this application;

[0042] Figure 2 This is a flowchart illustrating the battery life prediction method provided according to an embodiment of this application;

[0043] Figure 3 This is a framework diagram of the rolling time series index system provided according to the embodiments of this application;

[0044] Figure 4 This is a flowchart of the time-weighted AFM (Attentional Factorization Machine) model provided according to the embodiments of this application;

[0045] Figure 5 This is a flowchart illustrating the rolling time series network update prediction results according to an embodiment of this application.

[0046] Figure 6 This is a block diagram of a battery life prediction device provided according to an embodiment of this application;

[0047] Figure 7 This is an example diagram of a battery life prediction device provided according to an embodiment of this application;

[0048] Figure 8 This is a schematic diagram of the structure of a server provided according to an embodiment of this application. Detailed Implementation

[0049] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0050] The following describes a battery life prediction method, apparatus, server, and storage medium according to embodiments of this application with reference to the accompanying drawings. Addressing the problem mentioned in the background art that the prediction of battery life does not fully consider internal and external information during battery use, resulting in poor prediction accuracy, this application provides a method for battery life prediction based on a rolling time-series index system for battery degradation and a time-weighted AFM model. In this method, for historical vehicles with existing data reserves, a multi-dimensional battery degradation index system is established, including vehicle basic attributes, battery stability information, user driving information, and external operating condition information. All indicators are calculated by rolling the difference forward according to time stages, and then assigned low weight for earlier stages and high weight for later stages. The above independent variable indicators multiplied by the time weight are input into the AFM model for training. The trained model is used to predict the number of cycles corresponding to the battery life degradation node (SOH = 80%) for each vehicle that has been used for two years.

[0051] Specifically, Figure 1 This is a schematic flowchart of a battery life prediction method provided in an embodiment of this application.

[0052] like Figure 1 As shown, the battery life prediction method includes the following steps:

[0053] In step S101, vehicle data that has reached the battery life degradation node is obtained.

[0054] It is understood that the embodiments of this application can obtain vehicle data that has reached the battery life degradation node through the battery management system, vehicle control system and vehicle sensors.

[0055] In step S102, the basic attribute information of the vehicle and the internal and external information of the battery usage process are identified in the vehicle data. A first index dataset is established based on the basic attribute information, and a second index dataset is established based on the internal and external information of the battery usage process.

[0056] The battery usage process information includes one or more of the following: battery stability information, user driving information, and external operating condition information. The second indicator dataset includes one or more of the following: battery stability information indicator dataset, user driving information indicator dataset, and external operating condition information indicator dataset.

[0057] The first indicator dataset includes one or more of the following: vehicle model, battery manufacturer, battery material, battery production batch, standard capacity, and standard range.

[0058] It is understood that the embodiments of this application can establish first and second indicator datasets based on the vehicle's basic attribute information and internal and external information of battery usage, which facilitates subsequent prediction of battery life. These datasets include internal and external factors that affect battery life degradation, which can effectively improve the accuracy of prediction.

[0059] Optionally, in one embodiment of this application, a second indicator dataset is established based on information from inside and outside the battery usage process, including: obtaining the battery usage time; dividing the usage time into multiple time stages, and calculating the differences of all indicators in the information from inside and outside the battery usage process in the order of the time stages; and generating a second indicator dataset based on the differences of all indicators in all time stages.

[0060] It is understood that the embodiments of this application can calculate the differences of all indicators in the internal and external information of the battery usage process at different time stages based on the battery usage time, and generate a second indicator dataset based on the differences, which can effectively improve the data accuracy of the second indicator dataset.

[0061] In step S103, the indicators in the second indicator dataset are weighted according to the time stage, and the first prediction model is trained by combining the number of battery cycles corresponding to the battery life decay node. After training, the first indicator dataset and the first prediction model are associated.

[0062] In the second indicator dataset, the weight values ​​of the indicators can be set from small to large according to the order of time stages. For example, the time can be divided into 5 time stages with weights of 10%, 20%, 30%, etc., without specific limitations.

[0063] The first prediction model can be an AFM model, without specific limitations.

[0064] It is understood that the embodiments of this application can achieve the nonlinear trend of battery mechanism by assigning weights to the indicators in the second indicator dataset, and prepare for subsequent prediction of battery life by training the first prediction model.

[0065] In step S104, the first prediction model of the vehicle is matched according to the basic attribute information, the internal and external information of the vehicle's battery usage process is input into the first prediction model, and the first prediction cycle number corresponding to the battery life decay node of the vehicle is output.

[0066] It is understood that the embodiments of this application can match the first prediction model of the vehicle based on basic attribute information, and obtain the first prediction cycle number corresponding to the battery life decay node of the vehicle by inputting the internal and external information of the vehicle's battery usage process, which can effectively improve the accuracy of battery life prediction.

[0067] Optionally, in one embodiment of this application, after inputting the internal and external information of the vehicle's battery usage process into the first prediction model and outputting the first predicted cycle number corresponding to the battery life decay node of the vehicle, the method further includes: matching the vehicle's second prediction model based on basic attribute information; inputting the vehicle's battery health status (SOH) and actual cycle number into the second prediction model and outputting the second predicted cycle number corresponding to the battery life decay node of the vehicle; and using the second predicted cycle number to correct the first predicted cycle number to obtain the third predicted cycle number corresponding to the battery life decay node of the vehicle.

[0068] The second prediction model can be an LSTM (Long Short Term Memory Networks) model, without specific limitations.

[0069] It is understood that, in the embodiments of this application, after obtaining the first predicted cycle number corresponding to the battery life decay node of the vehicle, the second predicted cycle number corresponding to the battery life decay node of the vehicle can be obtained according to the second predicted cycle number, and the first predicted cycle number can be corrected according to the second predicted cycle number to obtain the third predicted cycle number, which can effectively improve the accuracy of the predicted cycle number.

[0070] Specifically, this application provides a method for updating battery life prediction values ​​using a rolling time-series network. The method includes the following: considering that there are many complex influencing factors during actual vehicle use, which can cause changes in the slope of the battery life degradation curve, each charging behavior can be used as the basic granularity, with SOH as the independent variable and the number of cycles as the dependent variable. The trend of the number of cycles corresponding to battery life degradation is tracked, and batteries from different manufacturers / materials / nominal specifications / operating conditions / regions are distinguished. The number of cycles is input into the LSTM model for training to obtain the number of cycles corresponding to the battery life degradation point. This result is used to update the prediction results obtained in the AFM model.

[0071] Optionally, in one embodiment of this application, the third predicted cycle number corresponding to the battery life degradation node of the vehicle is obtained by correcting the first predicted cycle number using the second predicted cycle number, including: identifying the rated cycle number of the battery in the basic attribute information; and calculating the third predicted cycle number based on the rated cycle number, the first predicted cycle number, and the second predicted cycle number.

[0072] It is understood that the embodiments of this application can calculate the third predicted cycle number based on the battery's rated cycle number, the first predicted cycle number, and the second predicted cycle number in the basic attribute information, which can effectively improve the accuracy of the predicted cycle number.

[0073] Optionally, in one embodiment of this application, the construction process of the second prediction model includes: acquiring training data, wherein the training data includes SOH and the number of iterations; constructing a nonlinear relationship curve between SOH and the number of iterations based on the training data, and clustering the nonlinear relationship curve based on basic attribute information; establishing a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, and associating the first index dataset with the second prediction model.

[0074] It is understood that the embodiments of this application can construct nonlinear relationship curves based on training data and perform clustering, and then establish a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, which can effectively realize the real-time updating of the prediction cycle number of vehicles.

[0075] According to the battery life prediction method proposed in this application, first and second indicator datasets can be established based on the basic attribute information of the vehicle and the internal and external information of the battery usage process. The first prediction model is trained by combining the number of battery cycles corresponding to the battery life decay node. By constructing a rolling time series indicator system to cover multiple factors inside and outside the battery, the model can be ensured to absorb as many dimensions of information as possible. In addition, the indicators in the second indicator dataset are weighted according to the time stage, which can meet the nonlinear trend of the battery mechanism, thereby effectively improving the accuracy of battery life prediction.

[0076] Based on the above embodiments, this application also proposes a battery life prediction method. The above embodiments are applied to offline states, while this embodiment is applied to online states. This embodiment and the previous embodiment have different focuses in their descriptions, and steps not described in each embodiment can be referred to each other.

[0077] like Figure 2 As shown, the battery life prediction method includes the following steps:

[0078] In step S201, the vehicle's basic attribute information and internal and external information about battery usage are obtained.

[0079] In step S202, the first prediction model of the vehicle is matched according to the basic attribute information. The internal and external information of the vehicle's battery usage process is input into the first prediction model, and the first prediction cycle number corresponding to the battery life decay node of the vehicle is output. The first prediction model is trained based on vehicle data that has reached the battery life decay node.

[0080] It is understood that the embodiments of this application can match the vehicle's first prediction model based on the vehicle's basic attribute information, and obtain the first prediction cycle number corresponding to the battery life decay node of the vehicle by inputting internal and external information of the battery usage process, which can effectively improve the accuracy of the prediction cycle number.

[0081] The battery life prediction method proposed in this application will be further illustrated below through another embodiment. It should be noted that the embodiments of this application are implemented based on the technical solution of this application, but the protection scope of this application is not limited to the following embodiments. In particular, this application proposes a battery life prediction method based on a rolling time series joint model, which includes the following:

[0082] First, in order to encompass both internal and external factors affecting battery life degradation, embodiments of this application provide a method for establishing a rolling time-series index system for battery degradation, such as... Figure 3 As shown, the method includes: for C that has reached the battery life degradation node (SOH = 80%) deg For each vehicle, the system obtains battery data from the battery management system for the two years prior to its deployment, driving data from the vehicle control system for the two years prior to its deployment, and external environmental data from the vehicle sensors for the two years prior to its deployment, and stores all of this data in the database.

[0083] The calculation of internal and external multi-dimensional battery metrics is based on a window format that scrolls backward through time stages. The data sources for these metrics all originate from the aforementioned database and specifically include four aspects: vehicle basic attributes D. std This includes vehicle model, battery manufacturer, battery materials, battery production batch, standard capacity, and standard range. This indicator uses the vehicle's initial factory data and does not require rolling calculations over time periods. Battery stability information D bat This includes the differences in battery capacity decay rate, voltage consistency, temperature difference, internal resistance consistency, and constant current ratio consistency between the first time stage T1 and the second time stage T2, as well as the aforementioned indicators between the second time stage T2 and the third time stage T3. These are calculated sequentially from time stage to time stage, and the user driving information D... dri This includes the user's speed consistency, acceleration consistency, maximum braking, maximum acceleration, and sustained speed in the first time phase T1, and the aforementioned indicators in the second time phase T2, calculated sequentially by time phase, along with external operating condition information D. out This includes the user's long-term road conditions in the first time period T1, such as highways or urban areas, long-term temperature zones, and long-term environmental conditions, such as flat or undulating terrain, as well as the aforementioned indicators for the user in the second time period T2, which are calculated by rolling forward according to the time period.

[0084] Secondly, in order to predict the battery degradation node C deg This application provides a method for establishing a time-weighted feature AFM model, such as... Figure 4 As shown, the method includes: using battery degradation node C deg The corresponding number of loops, Cyc degAs the dependent variable, the above indicators are used as independent variables, and time weights are assigned according to T1, T2, T3, etc. The specific weight values ​​can be set from low to high according to the order of the time stages. For example, it can be divided into 10 time stages with weights of 10%, 20%, 30%, etc.

[0085] In this process, after removing weakly correlated independent variables through correlation analysis, a high-dimensional feature set of time-weighted multiplication of independent variables is input into the AFM model. For example, the battery stability information input value is... For example, the user's driving information input value is For each new car that reaches its two-year lifespan, the number of cycles (Cyc_afm) corresponding to the battery life degradation node can be predicted using the model described above. deg .

[0086] In addition, to track the battery degradation trend of each new vehicle after a two-year period, and to achieve real-time updates of the vehicle's predicted cycle count (Cyc), deg This application provides a method for establishing a rolling time series network to update prediction results, such as... Figure 5 As shown, the method includes: obtaining the State of Health (SOH) and cycle count corresponding to the data uploaded by each vehicle during each charging from the battery management system; constructing a nonlinear relationship curve between SOH and cycle count for different manufacturers / materials / nominal values / operating conditions / regions; clustering the curves; and establishing several rolling time-series network update models based on the number of clusters. The cycle count is used as the dependent variable in the LSTM time-series model of the cluster type to which the vehicle belongs, thereby predicting the battery life degradation node C for each vehicle. deg The corresponding number of loops Cyc_lstm deg This result is then used to update the predictions obtained in the AFM model. The specific formula is as follows: Latest cycle number Cyc_new deg = (1 - current cycle count / nominal cycle count of battery sub-type × 100%) × predicted cycle count obtained from AFM model Cyc_afm deg +Current cycle count / Nominal cycle count for sub-type batteries × 100% × Predicted cycle count from LSTM model (Cyc_lstm) deg .

[0087] In summary, firstly, by constructing a rolling time-series index system that covers multiple factors both inside and outside the battery, the model can absorb as much dimensional information as possible. Furthermore, by setting time weights, the nonlinear trend of the battery mechanism can be satisfied. Secondly, actual business applications require both foresight and accuracy in battery life prediction. Through the dual-model joint tracking method for life changes described in this application, life decay nodes can be predicted in the early stages of vehicle use. The node data can be continuously improved based on changes in the actual usage of the vehicle, ensuring that the data has accuracy over time.

[0088] Next, the battery life prediction device according to the embodiments of this application is described with reference to the accompanying drawings.

[0089] Figure 6 This is a block diagram of a battery life prediction device according to an embodiment of this application.

[0090] like Figure 6 As shown, the battery life prediction device 10 includes: a first acquisition module 101, an establishment module 102, a training module 103, and a matching module 104.

[0091] The system comprises the following modules: a first acquisition module 101, used to acquire vehicle data that has reached the battery life degradation node; an establishment module 102, used to identify the vehicle's basic attribute information and internal and external battery usage information in the vehicle data, establish a first indicator dataset based on the basic attribute information, and establish a second indicator dataset based on the internal and external battery usage information; a training module 103, used to assign weights to the indicators in the second indicator dataset according to time stages, and train a first prediction model in combination with the battery cycle count corresponding to the battery life degradation node, and associate the first indicator dataset with the first prediction model after training; and a matching module 104, used to match the vehicle's first prediction model based on the basic attribute information, input the vehicle's internal and external battery usage information into the first prediction model, and output the first prediction cycle count corresponding to the vehicle's battery life degradation node.

[0092] Optionally, in one embodiment of this application, the internal and external information during battery use includes one or more of battery stability information, user driving information, and external operating condition information, and the second indicator dataset includes one or more of the indicator datasets of battery stability information, user driving information, and external operating condition information.

[0093] Optionally, in one embodiment of this application, the establishment module 102 is further configured to: obtain the battery usage time; divide the usage time into multiple time stages, and calculate the differences of all indicators in the internal and external information of the battery usage process in the order of the time stages; and generate a second indicator dataset based on the differences of all indicators in all time stages.

[0094] Optionally, in one embodiment of this application, the weight values ​​of the indicators in the second indicator dataset are set from small to large according to the chronological order of the time stages.

[0095] Optionally, in one embodiment of this application, the apparatus 10 further includes a correction module.

[0096] The correction module is used to input the internal and external information of the vehicle's battery usage process into the first prediction model and output the first prediction cycle number corresponding to the battery life decay node of the vehicle; then, it matches the second prediction model of the vehicle based on the basic attribute information; inputs the vehicle's battery health status (SOH) and actual cycle number into the second prediction model and outputs the second prediction cycle number corresponding to the battery life decay node of the vehicle; and corrects the first prediction cycle number using the second prediction cycle number to obtain the third prediction cycle number corresponding to the battery life decay node of the vehicle.

[0097] Optionally, in one embodiment of this application, the correction module is further configured to: identify the rated cycle number of the battery in the basic attribute information; and calculate the third predicted cycle number based on the rated cycle number, the first predicted cycle number, and the second predicted cycle number.

[0098] Optionally, in one embodiment of this application, the correction module can also be used to: acquire training data, wherein the training data includes SOH and the number of iterations; construct a nonlinear relationship curve between SOH and the number of iterations based on the training data, and cluster the nonlinear relationship curve based on basic attribute information; establish a corresponding second prediction model based on the nonlinear relationship curves of the same clustering type, and associate the first index dataset with the second prediction model.

[0099] It should be noted that the foregoing explanation of the battery life prediction method embodiment also applies to the battery life prediction device of this embodiment, and will not be repeated here.

[0100] According to the battery life prediction device proposed in this application, a first and second indicator dataset can be established based on the basic attribute information of the vehicle and the internal and external information of the battery usage process. The first prediction model is trained by combining the number of battery cycles corresponding to the battery life decay node. By constructing a rolling time series indicator system to cover multiple factors inside and outside the battery, the model can be ensured to absorb as many dimensions of information as possible. In addition, the indicators in the second indicator dataset are weighted according to the time stage, which can meet the nonlinear trend of the battery mechanism, thereby effectively improving the accuracy of battery life prediction.

[0101] Based on the above embodiments, this application also proposes a battery life prediction device. The above embodiments are applied to offline states, while this embodiment is applied to online states. This embodiment and the previous embodiment have different focuses in their descriptions, and steps not described in each embodiment can be referred to each other.

[0102] like Figure 7 As shown, the apparatus 20 in this embodiment includes a second acquisition module 201 and a prediction module 202.

[0103] The second acquisition module 201 is used to acquire the vehicle's basic attribute information and internal and external information of battery usage process; the prediction module 202 is used to match the vehicle's first prediction model according to the basic attribute information, input the vehicle's internal and external information of battery usage process into the first prediction model, and output the first prediction cycle number corresponding to the battery life decay node of the vehicle, wherein the first prediction model is trained based on vehicle data that has reached the battery life decay node.

[0104] It should be noted that the foregoing explanation of the battery life prediction method embodiment also applies to the battery life prediction device of this embodiment, and will not be repeated here.

[0105] According to the battery life prediction device proposed in this application, a first and second indicator dataset can be established based on the basic attribute information of the vehicle and the internal and external information of the battery usage process. The first prediction model is trained by combining the number of battery cycles corresponding to the battery life decay node. By constructing a rolling time series indicator system to cover multiple factors inside and outside the battery, the model can be ensured to absorb as many dimensions of information as possible. In addition, the indicators in the second indicator dataset are weighted according to the time stage, which can meet the nonlinear trend of the battery mechanism, thereby effectively improving the accuracy of battery life prediction.

[0106] Figure 8 A schematic diagram of the structure of a server provided in an embodiment of this application. The server may include:

[0107] The memory 801, the processor 802, and the computer program stored on the memory 801 and capable of running on the processor 802.

[0108] When the processor 802 executes the program, it implements the battery life prediction method provided in the above embodiments.

[0109] Furthermore, the server also includes:

[0110] Communication interface 803 is used for communication between memory 801 and processor 802.

[0111] The memory 801 is used to store computer programs that can run on the processor 802.

[0112] The memory 801 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0113] If the memory 801, processor 802, and communication interface 803 are implemented independently, then the communication interface 803, memory 801, and processor 802 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0114] Optionally, in a specific implementation, if the memory 801, processor 802, and communication interface 803 are integrated on a single chip, then the memory 801, processor 802, and communication interface 803 can communicate with each other through an internal interface.

[0115] The processor 802 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0116] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the battery life prediction method described above.

[0117] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0118] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0119] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0120] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0121] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium. When executed, the program includes one or a combination of the steps of the method embodiments.

[0122] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for predicting battery life, characterized in that, Includes the following steps: Acquire vehicle data that has reached the battery life degradation point; Identify the vehicle's basic attribute information and battery usage process internal and external information in the vehicle data, establish a first indicator dataset based on the basic attribute information, and establish a second indicator dataset based on the battery usage process internal and external information; Weights are assigned to the indicators in the second indicator dataset according to the time stage, and the first prediction model is trained in combination with the number of battery cycles corresponding to the battery life decay node. After training, the first indicator dataset and the first prediction model are associated. The vehicle's first prediction model is matched based on the basic attribute information. The internal and external information of the vehicle's battery usage process is input into the first prediction model, and the first prediction cycle number corresponding to the battery life decay node of the vehicle is output. The battery usage process internal and external information includes one or more of battery stability information, user driving information and external operating condition information, and the second indicator dataset includes one or more of the indicator datasets of battery stability information, user driving information and external operating condition information. The step of establishing a second indicator dataset based on internal and external information during battery usage includes: Get battery usage time; The usage time is divided into multiple time stages, and the differences of all indicators in the internal and external information of the battery usage process are calculated in the order of the time stages. The second indicator dataset is generated based on the differences between all indicators across all time periods.

2. The method according to claim 1, characterized in that, The weight values ​​of the indicators in the second indicator dataset are set from smallest to largest according to the chronological order of the time periods.

3. The method according to claim 1, characterized in that, After inputting internal and external information about the vehicle's battery usage process into the first prediction model and outputting the first prediction cycle number corresponding to the battery life degradation node of the vehicle, the method further includes: The second prediction model for the vehicle is matched based on the basic attribute information; The battery health status (SOH) and actual cycle count of the vehicle are input into the second prediction model, and the second predicted cycle count corresponding to the battery life degradation node of the vehicle is output. The first prediction cycle number is corrected using the second prediction cycle number to obtain the third prediction cycle number corresponding to the battery life degradation node of the vehicle.

4. The method according to claim 3, characterized in that, The step of correcting the first prediction cycle number using the second prediction cycle number to obtain the third prediction cycle number corresponding to the battery life degradation node of the vehicle includes: Identify the rated cycle count of the battery from the basic attribute information; The third predicted cycle number is calculated based on the rated cycle number, the first predicted cycle number, and the second predicted cycle number.

5. The method according to claim 3, characterized in that, The construction process of the second prediction model includes: Acquire training data, wherein the training data includes SOH and the number of iterations; A non-linear relationship curve between SOH and the number of cycles is constructed based on the training data, and the non-linear relationship curve is clustered based on the basic attribute information; A second prediction model is established based on the nonlinear relationship curves of the same clustering type, and the first index dataset is associated with the second prediction model.

6. A method for predicting battery life, characterized in that, Includes the following steps: Obtain basic vehicle attribute information and internal and external information regarding battery usage; The first prediction model is trained based on the basic attribute information of the vehicle according to the method described in claim 1. The internal and external information of the battery usage process of the vehicle is input into the first prediction model, and the first prediction cycle number corresponding to the battery life decay node of the vehicle is output. The first prediction model is trained based on vehicle data that has reached the battery life decay node.

7. A battery life prediction device, characterized in that, To implement the method according to any one of claims 1-5, comprising: The first acquisition module is used to acquire vehicle data that has reached the battery life degradation node; A module is established to identify the vehicle's basic attribute information and battery usage process internal and external information in the vehicle data, establish a first indicator dataset based on the basic attribute information, and establish a second indicator dataset based on the battery usage process internal and external information. The training module is used to assign weights to the indicators in the second indicator dataset according to the time stage, and train the first prediction model in combination with the number of battery cycles corresponding to the battery life decay node. After the training is completed, the first indicator dataset is associated with the first prediction model. The matching module is used to match the vehicle's first prediction model based on the basic attribute information, input the vehicle's internal and external battery usage information into the first prediction model, and output the first prediction cycle number corresponding to the battery life decay node of the vehicle.

8. A battery life prediction device, characterized in that, To implement the method of claim 6, comprising: The second acquisition module is used to acquire basic vehicle attribute information and internal and external information about battery usage. The prediction module is used to match the vehicle's first prediction model based on the basic attribute information, input the vehicle's internal and external battery usage information into the first prediction model, and output the first prediction cycle number corresponding to the battery life decay node of the vehicle. The first prediction model is trained based on vehicle data that has reached the battery life decay node.

9. A server, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the battery life prediction method as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the battery life prediction method as described in any one of claims 1-6.

Images