Battery deterioration prediction system
By constructing a battery degradation prediction system and using a database and dataset selection unit to select a suitable dataset for model learning, the problem of low accuracy in predicting battery degradation status in existing technologies is solved, and higher accuracy in battery degradation prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TOYOTA JIDOSHA KK
- Filing Date
- 2025-07-25
- Publication Date
- 2026-05-12
AI Technical Summary
Existing machine learning-based battery degradation prediction models may have lower accuracy, making it difficult to accurately predict the future degree of battery degradation.
By constructing a battery degradation prediction system, a database is used to store secondary battery data from multiple vehicles. A dataset selection unit is employed to select a suitable dataset for model learning, including data preprocessing and cluster analysis, to build a high-precision battery degradation prediction model.
It improves the accuracy of battery degradation prediction, reduces the impact of differences in vehicle characteristics, and enhances the model's prediction accuracy.
Smart Images
Figure CN122017638A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of a battery degradation prediction system for predicting the degradation degree of secondary batteries. Background Technology
[0002] As such a system, for example, the following system has been proposed: a learning prediction model, which, based on the learned prediction model, estimates indicators related to the degradation state of the battery (see Japanese Patent Application Publication No. 2023-51009), the prediction model estimates indicators related to the degradation state of the battery based on learning data. Summary of the Invention
[0003] In the technology described in Japanese Patent Application Publication No. 2023-51009, indicators related to the degradation state of a battery are estimated based on a predictive model learned through machine learning. However, the accuracy of the estimation may decrease depending on the learning data used in the machine learning.
[0004] The present invention was made in view of the above-mentioned problems, and its objective is to provide a battery degradation prediction system that can improve the estimation accuracy.
[0005] One aspect of the present invention provides a battery degradation prediction system comprising: a database storing degradation levels of secondary batteries for driving in multiple vehicles; and a dataset selection unit that changes the dataset used in learning a model for predicting the future degradation levels of the secondary batteries in each of the multiple vehicles based on the storage status of the database. Attached Figure Description
[0006] The features, advantages, and technical and industrial significance of exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, in which the same reference numerals denote the same elements, and wherein:
[0007] Figure 1 This is a schematic diagram showing the structure of the battery degradation prediction system according to the implementation method;
[0008] Figure 2 This is a flowchart illustrating the operation of the battery degradation prediction system in an implementation method.
[0009] Figure 3 This is a flowchart illustrating the methods used to determine the data used in learning; and
[0010] Figure 4 This is a flowchart representing the model construction method. Detailed Implementation
[0011] Reference Figures 1 to 4 The implementation method of the battery degradation prediction system is described. Figure 1The battery degradation prediction system 10 includes a database 11, a dataset selection unit 12, and a model learning unit 13. For example, the battery degradation prediction system 10 can be implemented by a server. It should be noted that the battery degradation prediction system 10 can be implemented by a single server or by multiple servers. The server can be a cloud server.
[0012] The battery degradation prediction system 10 is configured to communicate bidirectionally with multiple vehicles V1, V2, ..., Vn via the Internet. Each of the vehicles V1, V2, ..., Vn is equipped with a secondary battery B1, B2, ..., Bn for propulsion. The vehicles V1, V2, ..., Vn may include at least one of battery electric vehicles, plug-in hybrid electric vehicles, hybrid electric vehicles, and fuel cell electric vehicles. The secondary batteries B1, B2, ..., Bn may include at least one of lithium-ion batteries, nickel-metal hydride batteries, and all-solid-state batteries.
[0013] Multiple vehicles V1, V2, ..., Vn periodically send data containing the State of Health (SOH) of the secondary batteries to the battery degradation prediction system 10. This data may include specification information, timestamps, vehicle and battery-related characteristics, and degradation-related information. For example, specification information may include the Vehicle Identification Number (VIN), battery serial number, battery model, vehicle model, destination, and design change version. Degradation-related information may include degradation level, number of days elapsed since Line Off (L / O) (i.e., production completion date), mileage, cumulative parking time, SOC (State of Charge) history and battery temperature history during ignition ON, and SOC history and battery temperature history during ignition OFF. It should be noted that ignition ON may include at least one of the following: the vehicle is in motion, in power delivery mode, or charging. Ignition OFF may include at least one of the following: parked or in transit. Multiple data points sent from multiple vehicles V1, V2, ..., Vn are stored in database 11. As a result, the degradation degree of secondary batteries B1, B2, ..., Bn is stored in database 11.
[0014] The battery degradation prediction system 10 uses data on the degradation levels of secondary batteries stored in the database 11 to learn a model that predicts the future degradation level of the secondary batteries. (Refer to...) Figures 2 to 4 The operation of the battery degradation prediction system 10 is explained. The aforementioned model can, for example, apply multiple regression models, multivariate machine learning models, and DNN (Deep Neural Network).
[0015] exist Figure 2 In the battery degradation prediction system 10, the dataset selection unit 12 determines the dataset stored in the database 11 that will be used in learning the model for predicting the future degradation of a vehicle's secondary battery (step S10). See also... Figure 3 The processing of step S10 will be explained below. Hereinafter, the above-mentioned vehicle will be referred to as vehicle V1.
[0016] The dataset selection unit 12 can set the values of thresholds A, B, and C (described later) based on data related to vehicle V1 stored in the database 11. For example, the values of thresholds A, B, and C can vary depending on the battery model, vehicle model, destination, and design change version. Alternatively, the dataset selection unit 12 can set the values of thresholds A, B, and C based on a table specifying the relationship between battery model, vehicle model, destination, design change version, and thresholds A, B, and C. It should be noted that the values of thresholds A, B, and C can also be predetermined fixed values.
[0017] exist Figure 3 In step S101, the dataset selection unit 12 determines whether the number N1 of data related to vehicle V1 stored in database 11 is greater than a threshold A. If, in step S101, it is determined that the number N1 is greater than the threshold A (step S101: YES), the dataset selection unit 12 decides to use only the multiple data related to vehicle V1 in the learning of the model predicting the future degradation degree of the secondary battery B1 of vehicle V1 (step S102). In this case, the dataset used for learning includes only multiple data containing the degradation degree of the secondary battery B1 of vehicle V1.
[0018] In step S101, if it is determined that the quantity N1 is below threshold A (step S101: No), the dataset selection unit 12 determines whether the total number N2 of the data related to vehicle V1 and the data of other vehicles of the same type as vehicle V1 stored in the database 11 is greater than threshold B. In step S103, if it is determined that the total number N2 is greater than threshold B (step S103: YES), the dataset selection unit 12 decides to use the data related to vehicle V1 and the data of other vehicles of the same type as vehicle V1 in the learning of a model predicting the future degradation degree of the secondary battery B1 of vehicle V1 (step S104). In this case, the dataset used in the learning includes more than one set of data containing the degradation degree of the secondary battery B1 of vehicle V1, and more than one set of data containing the degradation degree of the secondary batteries of other vehicles of the same type as vehicle V1.
[0019] In step S103, if the total N2 is determined to be below threshold B (step S103: No), the dataset selection unit 12 determines whether the total N3 of the data related to vehicle V1 and the data related to other vehicles equipped with secondary batteries of the same type as vehicle V1's secondary battery B1 stored in the database 11 is greater than threshold C (step S105). In step S105, if the total N3 is determined to be greater than threshold C (step S105: Yes), the dataset selection unit 12 decides to use the data related to vehicle V1 and the data related to other vehicles equipped with secondary batteries of the same type as vehicle V1's secondary battery B1 in the learning of a model predicting the future degradation degree of vehicle V1's secondary battery B1 (step S106). In this case, the dataset used in the learning includes more than one set of data containing the degradation degree of vehicle V1's secondary battery B1 and more than one set of data containing the degradation degree of secondary batteries of other vehicles equipped with secondary batteries of the same type as vehicle V1's secondary battery B1.
[0020] In step S105, if the total N3 is determined to be below the threshold C (step S105: No), the dataset selection unit 12 decides to use the data related to vehicle V1 and the data related to other vehicles stored in the database 11 in the learning of the model predicting the future degradation of the secondary battery B1 of vehicle V1 (step S107). In this case, the other vehicles may include vehicles of a different type than vehicle V1. Additionally, the other vehicles may include vehicles equipped with a secondary battery of a different type than the secondary battery B1 of vehicle V1. It should be noted that the dataset selection unit 12 may also, for example, decide to use the data related to vehicle V1 and the data of other vehicles with the same destination as vehicle V1 in the learning of the aforementioned model. Furthermore, the dataset selection unit 12 may also, for example, decide to use the data related to vehicle V1 and the data of other vehicles with the same purpose as vehicle V1 (commercial use, etc.) in the learning of the aforementioned model.
[0021] return Figure 2 After step S10, the model learning unit 13 of the battery degradation prediction system 10 uses the dataset determined by the processing in step S10 to construct a model (e.g., a model predicting the future degradation of the secondary battery B1 of vehicle V1) (step S20). See also... Figure 4 The processing of step S20 is explained.
[0022] exist Figure 4 In this process, the model learning unit 13 excludes exceptional data from the data contained in the dataset (step S201). This configuration improves data quality. Exceptional data can include data related to vehicles with replaced drive batteries, data related to vehicles where the degradation characteristics of the drive battery have changed due to software design changes, data with large deviations in degradation degree, and data with low measurement accuracy due to certain measurement failures. It should be noted that the determination of whether data is exceptional can be achieved using, for example, rule-based determination or unsupervised learning algorithms such as k-means.
[0023] Next, the model learning unit 13 sets the ground truth value for the degradation degree (step S202). It should be noted that various existing methods can be used to set the ground truth value for the degradation degree. Therefore, a detailed explanation of the method for setting the ground truth value for the degradation degree is omitted. Next, the model learning unit 13 excludes data from the dataset from which a ground truth value for the degradation degree cannot be set (step S203). If configured in this way, the quality of the data can be improved.
[0024] Next, the model learning unit 13 clusters the data contained in the dataset for each vehicle based on the degradation tendency (step S204). In the process of step S204, the model learning unit 13 may also sample only the data with properties that are similar to those of the predicted object (e.g., the secondary battery B1 of vehicle V1).
[0025] Next, the model learning unit 13 uses the dataset to construct a model predicting the future degradation of the secondary battery (e.g., a model predicting the future degradation of the secondary battery B1 of vehicle V1) (step S205). Next, the model learning unit 13 evaluates the accuracy of the model constructed in step S205 (step S206). It should be noted that various existing methods can be applied to the model construction method and the method for evaluating the accuracy of the constructed model. Therefore, detailed descriptions of the model construction method and the method for evaluating the accuracy of the constructed model are omitted.
[0026] It should be noted that when the dataset contains data related to other vehicles in addition to data related to a single vehicle (e.g., vehicle V1), the impact of data related to other vehicles on the model may be greater than the impact of data related to a single vehicle. Therefore, the model learning unit 13 can perform subsampling of the dataset, making the weights of data related to a single vehicle greater than the weights of data related to other vehicles. For example, the weighting of data can be implemented arbitrarily in modeling or learned in a way that is adjusted in machine learning or statistical models. Regarding data subsampling, it can be performed such that more data related to other vehicles with characteristics similar to those of a single vehicle is retained, while less data related to other vehicles with characteristics different from those of a single vehicle is retained. For example, the model learning unit 13 can also perform random subsampling so that the ratio of data related to a single vehicle (e.g., vehicle V1) to data related to other vehicles is a:b.
[0027] return Figure 1 After step S20, the battery degradation prediction system 10 determines whether the accuracy of the constructed model is greater than a predetermined threshold (step S30). In the processing of step S30, if it is determined that the accuracy of the constructed model is greater than the predetermined threshold (step S30: YES), the battery degradation prediction system 10 decides to adopt the constructed model (step S40). In this case, the battery degradation prediction system 10 can use the constructed model to predict the future degradation of a vehicle's secondary battery (e.g., secondary battery B1 of vehicle V1).
[0028] In step S30, if the accuracy of the constructed model is determined to be below a predetermined threshold (step S30: NO), the battery degradation prediction system 10 decides not to use the constructed model (step S50). In this case, the battery degradation prediction system 10 can perform another step... Figure 2 The actions are shown in the flowchart.
[0029] Technical effect
[0030] When a sufficient amount of data related to a single vehicle (i.e., data on the degradation degree of a vehicle's secondary battery) is available, it is desirable to construct a model predicting future degradation using only the data related to that single vehicle. This is because if data related to other vehicles is used in addition to the data related to that single vehicle, the differences in characteristics between the two vehicles may affect the model. In this embodiment, the processing in step S101 (refer to...) Figure 3 In this approach, if the number N1 is determined to be greater than the threshold A, a dataset containing multiple data points related to only one vehicle (e.g., vehicle V1) is used in the model's learning process. In this case, the constructed model is not affected by the differences in characteristics between one vehicle and other vehicles, thus improving the accuracy of degradation prediction can be expected.
[0031] When the amount of data related to a single vehicle is insufficient, utilizing data related to other vehicles can increase the amount of data available for the model's learning. However, as mentioned above, the differences in characteristics between one vehicle and other vehicles can potentially affect the model. Therefore, in this embodiment, data related to other vehicles to be utilized is selected based on the available data (see [reference]). Figure 3 (Flowchart). By constructing it in this way, it is possible to suppress the degradation of the data used in the learning of the above model and to increase the amount of data. In this case, it is also possible to expect an improvement in the prediction accuracy of the degradation degree.
[0032] The invention described below is a method of invention derived from the embodiments described above.
[0033] One aspect of the battery degradation prediction system includes: a database storing the degradation levels of secondary batteries used for driving multiple vehicles; and a dataset selection unit that changes the dataset used in learning a model to predict the future degradation levels of the secondary batteries in each of the multiple vehicles based on the storage status of the database. In the above embodiment, "dataset selection unit 12" is an example of "dataset selection unit".
[0034] In one example of this battery degradation prediction system, the dataset selection unit may select a dataset containing multiple degradation values of the secondary batteries of the one vehicle stored in the database, based on the storage status of the database, as the dataset used in learning a model to predict the future degradation value of the secondary batteries of one of the multiple vehicles.
[0035] In another example of this battery degradation prediction system, the dataset selection unit may select a dataset containing multiple degradation values of the secondary batteries of the vehicle and vehicles of the same type stored in the database, based on the storage status of the database, as the dataset used in learning a model to predict the future degradation value of the secondary battery of one of the vehicles.
[0036] In another example of this battery degradation prediction system, the dataset selection unit may select a dataset containing multiple degradation values of the secondary batteries of the vehicle and secondary batteries of the same type as the secondary batteries of the vehicle stored in the database, based on the storage status of the database, as the dataset used in learning a model to predict the future degradation value of the secondary batteries of one of the multiple vehicles.
[0037] This invention is not limited to the embodiments described above, and can be appropriately modified without departing from the spirit or spirit of the invention as read in its entirety from the claims and specification. In addition, the battery degradation prediction system accompanying such modifications is also included within the scope of protection of this invention.
Claims
1. A battery degradation prediction system, wherein, This battery degradation prediction system has the following features: A database storing the degradation levels of secondary batteries used in the drive systems of multiple vehicles; and The dataset selection unit changes the dataset used in learning the model to predict the future degradation of the secondary batteries of the multiple vehicles, based on the storage status of the database.
2. The battery degradation prediction system according to claim 1, wherein, The dataset selection unit selects a dataset containing multiple degradation degrees of the secondary battery of one of the multiple vehicles stored in the database, based on the storage status of the database, as the dataset to be used in learning a model that predicts the future degradation degree of the secondary battery of the one vehicle.
3. The battery degradation prediction system according to claim 1, wherein, The dataset selection unit selects a dataset containing multiple degradation levels of the secondary batteries of one of the plurality of vehicles stored in the database and vehicles of the same type as the one vehicle, based on the storage status of the database, as the dataset to be used in learning a model that predicts the future degradation level of the secondary batteries of the one vehicle.
4. The battery degradation prediction system according to claim 1, wherein, The dataset selection unit selects a dataset containing the secondary battery of one of the plurality of vehicles stored in the database and multiple degradation degrees of secondary batteries of the same type as the secondary battery of the vehicle, based on the storage status of the database, as the dataset to be used in learning a model that predicts the future degradation degree of the secondary battery of the vehicle.