A battery capacity prediction method and system
By constructing and training a battery capacity prediction model and using hypothesis testing methods to test the results, the problem of difficulty in efficiently calculating the actual capacity of a vehicle battery in the prior art is solved, and efficient and accurate battery capacity prediction is achieved.
Patent Information
- Application Number
- CN202411927430.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-25
- Publication Date
- 2025-04-04
AI Technical Summary
The prior art is difficult to efficiently and accurately calculate the actual capacity of large-volume vehicle batteries, and lacks fast and effective inspection methods.
By obtaining the vehicle's battery charging data, filtering the ideal charging data, extracting the charging feature set, and building and training a battery capacity prediction model, the prediction results are tested using hypothesis testing method.
It realizes efficient prediction of the actual capacity of the battery of the rechargeable vehicle, improves the efficiency and accuracy of the prediction, and improves the reliability of the battery capacity prediction model.
Smart Images
Figure CN119758134A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of automobile battery capacity prediction, and in particular to a battery capacity prediction method and system. Background Art
[0002] In recent years, as the sales of new energy vehicles continue to rise, the large-scale use of energy storage batteries and power batteries has become the main consensus in the future new energy field. During the use of the battery itself, irreversible physical and chemical processes will occur inside it. As the number of battery charge and discharge cycles increases, the battery's internal resistance will increase and its capacity will decrease.
[0003] Battery capacity degradation can cause problems such as reduced battery performance and shortened service life, which may further lead to performance degradation of corresponding electrical equipment or electrical equipment system failure. Therefore, in order to better deal with the potential impact of battery capacity degradation, it is particularly important to predict battery capacity. There are currently two main methods for calculating the actual capacity of vehicle batteries: 1. Detect the effective power of the battery's full charge and discharge cycle through offline instruments; 2. Calculate the effective power of the vehicle after it is fully charged after being discharged to an SOC of less than 20% and stationary for 2 hours through online data.
[0004] However, as the number of vehicles increases, the above-mentioned method for obtaining the actual capacity of vehicle batteries has high calculation costs and difficult to achieve calculation conditions, which makes it difficult to calculate the actual capacity of vehicle batteries in large quantities. In addition, there is a lack of effective verification methods for large amounts of calculation data, and it is impossible to quickly and effectively verify the calculation data. Summary of the invention
[0005] The present invention aims to provide a battery capacity prediction method and system to solve the above-mentioned technical problems, realize battery capacity prediction through a battery capacity prediction model, and test the battery capacity prediction results based on a hypothesis testing method, so as to effectively improve the efficiency and accuracy of battery capacity prediction.
[0006] In order to solve the above technical problems, the present invention provides a battery capacity prediction method, comprising the following steps:
[0007] Obtain battery charging data of all vehicles; filter ideal charging data based on the battery charging data; obtain an ideal charging feature set based on the ideal charging data, and build and train a battery capacity prediction model; extract a charging feature set based on the ideal charging feature set and the battery charging data; use the charging feature set as input to the battery capacity prediction model so that the battery capacity prediction model predicts the battery capacity of each vehicle, thereby obtaining a battery capacity prediction value for each vehicle; test the battery capacity prediction value for each vehicle based on a hypothesis testing method to obtain a battery capacity prediction result for the vehicle.
[0008] In the above scheme, the actual capacity of the battery of the charging vehicle can be predicted through the battery capacity prediction model, which has a high coverage rate; and the battery capacity prediction results are tested based on the hypothesis testing method, which effectively improves the efficiency and accuracy of the battery capacity prediction and enhances the reliability of the battery capacity prediction model.
[0009] It should be noted that the battery charging data of the vehicle may be online data. The charging feature set may be calculated and extracted through the online data, and the charging feature set may be input into the trained battery capacity prediction model to predict the actual capacity of the vehicle battery.
[0010] Furthermore, the filtering of ideal charging data based on battery charging data includes: filtering battery charging data that meets the following preset filtering conditions as ideal charging data: Condition 1, the vehicle needs to be discharged until the battery SOC is less than a preset first capacity threshold; Condition 2, the vehicle needs to be left stationary for a preset first time before charging; Condition 3, the vehicle needs to be charged until the battery SOC is greater than a preset second capacity threshold.
[0011] It should be noted that the first capacity threshold may be 20%, the first duration may be 2 hours, and the second capacity threshold may be 100%.
[0012] Furthermore, the method of obtaining an ideal charging feature set based on the ideal charging data, and constructing and training a battery capacity prediction model includes: obtaining an ideal charging feature set based on the ideal charging data; calculating the ideal battery capacity of the vehicle corresponding to the ideal charging data; constructing an initial prediction model, using the ideal charging feature set as the input of the initial prediction model, and using the ideal battery capacity as the label of the ideal charging feature, and then training the initial prediction model to obtain a battery capacity prediction model.
[0013] Furthermore, the initial prediction model includes a variety of nonlinear regression models such as a neural network model or a random forest model.
[0014] Furthermore, the extracting of the ideal charging feature set based on the ideal charging data includes: obtaining a number of charging features related to the battery capacity in the ideal charging data; calculating the correlation between the number of charging features and the battery capacity based on correlation analysis, and extracting the ideal charging feature set.
[0015] In the above scheme, the charging features with high correlation with the battery capacity can be screened out as the ideal charging feature set through correlation analysis. The ideal charging feature set is input and combined with the calculated ideal battery capacity to train the initial prediction model, which is beneficial to improve the accuracy of the battery capacity prediction model training.
[0016] It should be noted that the charging characteristics may include the mean, variance and range of the cumulative mileage, battery cell temperature, battery cell voltage difference or battery cell voltage to power increment fitting curve, etc. The correlation between different characteristics and the actual battery capacity can be calculated through correlation analysis to select the characteristics with high correlation.
[0017] Furthermore, the calculation of the ideal battery capacity of the vehicle corresponding to the ideal charging data includes: extracting the ideal charging vehicle corresponding to the ideal charging data; calculating the ideal battery capacity C of the ideal charging vehicle n ,have:
[0018]
[0019] Where: C t is the SOC at the end of charging; C0 is the SOC at the start of charging; C is the rated SOC of the battery; I is the charging current.
[0020] Further, the extracting the charging feature set based on the ideal charging feature set and the battery charging data includes: determining a feature type based on the ideal charging feature set; and extracting features from the battery charging data based on the feature type to obtain the charging feature set.
[0021] In the above scheme, determining the feature type based on the ideal charging feature set can ensure the consistency of the charging feature set with the ideal charging feature set, thereby ensuring the accuracy of the battery capacity prediction model prediction.
[0022] Furthermore, the battery capacity prediction value of each vehicle is tested based on the hypothesis testing method to obtain the battery capacity prediction result of the vehicle, including: taking the battery capacity prediction value of each vehicle as sample x, taking the ideal battery capacity of the vehicle corresponding to the ideal charging data as sample y, assuming that the mean of sample x is no different from the mean of sample y, then according to the central limit theorem, calculating the verification value Z, there is:
[0023]
[0024] Where: μ x represents the mean of sample x; μ y represents the mean of sample y; represents the variance of sample x; represents the variance of sample y;
[0025] If the verification value Z is less than the preset error value, the hypothesis is accepted, and it is considered that there is no difference between the mean of sample x and the mean of sample y, and the battery capacity prediction result of the vehicle is accurate.
[0026] In the above scheme, by calculating the mean and variance of the ideal battery capacity and the predicted battery capacity, the hypothesis test method is used to determine whether the two samples come from the same population. If the test is passed, it can be considered that the trained battery capacity prediction model has high accuracy; otherwise, it is considered that the trained battery capacity prediction model has low accuracy and needs to be retrained.
[0027] Furthermore, the preset error value is obtained from a preset query table based on a preset single significance level.
[0028] The above solution does not rely on offline equipment, and can directly establish a battery capacity prediction model through online data to predict the actual capacity of batteries for large-scale charging vehicles, thereby improving the efficiency of battery capacity prediction. In addition, by adopting the hypothesis testing method to verify the actual battery capacity prediction, the accuracy of multiple battery capacity prediction values can be tested without comparing samples and observed values, thus ensuring the reliability of the actual battery capacity prediction.
[0029] The present invention further provides a battery capacity prediction system, which is used to execute the above-mentioned battery capacity prediction method, comprising:
[0030] A data acquisition module is used to acquire battery charging data of all vehicles; a data screening module is used to screen ideal charging data based on battery charging data; a feature extraction module is used to acquire an ideal charging feature set based on the ideal charging data, and to extract a charging feature set based on the ideal charging feature set and the battery charging data; a model construction module is used to construct and train a battery capacity prediction model based on the ideal charging feature set; a capacity prediction module is used to use the charging feature set as an input to the battery capacity prediction model so that the battery capacity prediction model predicts the battery capacity of each vehicle, thereby obtaining a battery capacity prediction value for each vehicle; a prediction verification module is used to verify the battery capacity prediction value of each vehicle based on a hypothesis testing method, and obtain a battery capacity prediction result for the vehicle.
[0031] The system proposed in the above scheme has a simple architecture and is easy to build. It can quickly and effectively implement a battery capacity prediction method. The battery capacity prediction model can be used to predict the actual capacity of the battery of the charging vehicle, with a high coverage rate. The battery capacity prediction results are tested based on the hypothesis testing method, which effectively improves the efficiency and accuracy of the battery capacity prediction and enhances the reliability of the battery capacity prediction model. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 A schematic diagram of a battery capacity prediction method according to an embodiment of the present invention;
[0033] Figure 2 A battery capacity prediction system architecture diagram provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION
[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0035] See also Figure 1 This embodiment provides a battery capacity prediction method, comprising the following steps:
[0036] S1: Obtain battery charging data of all vehicles;
[0037] S2: Filter ideal charging data based on battery charging data;
[0038] S3: Obtain an ideal charging feature set based on the ideal charging data, and build and train a battery capacity prediction model;
[0039] S4: extracting a charging feature set based on an ideal charging feature set and battery charging data;
[0040] S5: using the charging feature set as an input of a battery capacity prediction model, so that the battery capacity prediction model predicts the battery capacity of each vehicle, and then obtaining a predicted value of the battery capacity of each vehicle;
[0041] S6: Testing the predicted value of the battery capacity of each vehicle based on the hypothesis testing method to obtain the predicted result of the battery capacity of the vehicle.
[0042] In the specific implementation process of this embodiment, the actual capacity of the battery of the charging vehicle can be predicted through the battery capacity prediction model, with a high coverage rate; and the battery capacity prediction result is tested based on the hypothesis testing method, which effectively improves the efficiency and accuracy of the battery capacity prediction and enhances the reliability of the battery capacity prediction model.
[0043] It should be noted that the battery charging data of the vehicle may be online data. The charging feature set may be calculated and extracted through the online data, and the charging feature set may be input into the trained battery capacity prediction model to predict the actual capacity of the vehicle battery.
[0044] Furthermore, the filtering of ideal charging data based on battery charging data includes: filtering battery charging data that meets the following preset filtering conditions as ideal charging data: Condition 1, the vehicle needs to be discharged until the battery SOC is less than a preset first capacity threshold; Condition 2, the vehicle needs to be left stationary for a preset first time before charging; Condition 3, the vehicle needs to be charged until the battery SOC is greater than a preset second capacity threshold.
[0045] It should be noted that the first capacity threshold may be 20%, the first duration may be 2 hours, and the second capacity threshold may be 100%.
[0046] Furthermore, the method of obtaining an ideal charging feature set based on the ideal charging data, and constructing and training a battery capacity prediction model includes: obtaining an ideal charging feature set based on the ideal charging data; calculating the ideal battery capacity of the vehicle corresponding to the ideal charging data; constructing an initial prediction model, using the ideal charging feature set as the input of the initial prediction model, and using the ideal battery capacity as the label of the ideal charging feature, and then training the initial prediction model to obtain a battery capacity prediction model.
[0047] Furthermore, the initial prediction model includes a variety of nonlinear regression models such as a neural network model or a random forest model.
[0048] Furthermore, the extracting of the ideal charging feature set based on the ideal charging data includes: obtaining a number of charging features related to the battery capacity in the ideal charging data; calculating the correlation between the number of charging features and the battery capacity based on correlation analysis, and extracting the ideal charging feature set.
[0049] In this embodiment, charging features with a high correlation with battery capacity can be screened out as an ideal charging feature set through correlation analysis. The ideal charging feature set is input and combined with the calculated ideal battery capacity to train the initial prediction model, which is beneficial to improving the accuracy of battery capacity prediction model training.
[0050] It should be noted that the charging characteristics may include the mean, variance and range of the cumulative mileage, battery cell temperature, battery cell voltage difference or battery cell voltage to power increment fitting curve, etc. The correlation between different characteristics and the actual battery capacity can be calculated through correlation analysis to select the characteristics with high correlation.
[0051] Furthermore, the calculation of the ideal battery capacity of the vehicle corresponding to the ideal charging data includes: extracting the ideal charging vehicle corresponding to the ideal charging data; calculating the ideal battery capacity C of the ideal charging vehicle n ,have:
[0052]
[0053] Where: C t is the SOC at the end of charging; C0 is the SOC at the start of charging; C is the rated SOC of the battery; I is the charging current.
[0054] Further, the extracting the charging feature set based on the ideal charging feature set and the battery charging data includes: determining a feature type based on the ideal charging feature set; and extracting features from the battery charging data based on the feature type to obtain the charging feature set.
[0055] In this embodiment, determining the feature type based on the ideal charging feature set can ensure the consistency of the charging feature set and the ideal charging feature set, thereby ensuring the accuracy of the battery capacity prediction model prediction.
[0056] Furthermore, the battery capacity prediction value of each vehicle is tested based on the hypothesis testing method to obtain the battery capacity prediction result of the vehicle, including: taking the battery capacity prediction value of each vehicle as sample x, taking the ideal battery capacity of the vehicle corresponding to the ideal charging data as sample y, assuming that the mean of sample x is no different from the mean of sample y, then according to the central limit theorem, calculating the verification value Z, there is:
[0057]
[0058] Where: μ x represents the mean of sample x; μ y represents the mean of sample y; represents the variance of sample x; represents the variance of sample y;
[0059] If the verification value Z is less than the preset error value, the hypothesis is accepted, and it is considered that there is no difference between the mean of sample x and the mean of sample y, and the battery capacity prediction result of the vehicle is accurate.
[0060] In this embodiment, by calculating the mean and variance of the ideal battery capacity and the battery capacity prediction value, the hypothesis test method is used to determine whether the two samples come from the same population. If the test is passed, it can be considered that the trained battery capacity prediction model has a high accuracy; otherwise, it is considered that the trained battery capacity prediction model has a low accuracy and needs to be retrained.
[0061] Furthermore, the preset error value is obtained from a preset query table based on a preset single significance level.
[0062] In one embodiment, the mean of sample x is calculated to be 101.9712, and the standard deviation is 19.3419; the mean of sample y is 98.7249, and the standard deviation is 3.0298, and the verification value Z=0.1654 can be calculated. By corresponding to Z at the single significance level of 0.05 * By comparing the values, we find that Z=0.1654<1.64, so the predicted actual capacity of the battery can be considered accurate.
[0063] The method provided in this embodiment does not rely on offline equipment, and can directly establish a battery capacity prediction model through online data to predict the actual capacity of batteries of large-scale charging vehicles, thereby improving the efficiency of battery capacity prediction. In addition, by adopting the hypothesis testing method to verify the actual battery capacity prediction, the accuracy of multiple battery capacity prediction values can be tested without comparing samples and observed values, thereby ensuring the reliability of the actual battery capacity prediction.
[0064] See also Figure 2 This embodiment provides a battery capacity prediction system, which is used to execute the above-mentioned battery capacity prediction method, including:
[0065] A data acquisition module is used to acquire battery charging data of all vehicles; a data screening module is used to screen ideal charging data based on battery charging data; a feature extraction module is used to acquire an ideal charging feature set based on the ideal charging data, and to extract a charging feature set based on the ideal charging feature set and the battery charging data; a model construction module is used to construct and train a battery capacity prediction model based on the ideal charging feature set; a capacity prediction module is used to use the charging feature set as an input to the battery capacity prediction model so that the battery capacity prediction model predicts the battery capacity of each vehicle, thereby obtaining a battery capacity prediction value for each vehicle; a prediction verification module is used to verify the battery capacity prediction value of each vehicle based on a hypothesis testing method, and obtain a battery capacity prediction result for the vehicle.
[0066] The system proposed in this embodiment has a simple architecture and is easy to build. It can quickly and effectively implement a battery capacity prediction method. The battery capacity prediction model can be used to predict the actual capacity of the battery of a charging vehicle, with a high coverage rate. The battery capacity prediction results are tested based on the hypothesis testing method, which effectively improves the efficiency and accuracy of the battery capacity prediction and enhances the reliability of the battery capacity prediction model.
[0067] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A battery capacity prediction method, characterized in that: The following steps are involved: Get battery charging data for all vehicles; Filter ideal charging data based on battery charging data; Obtain an ideal charging feature set based on ideal charging data, and build and train a battery capacity prediction model; Extracting a charging feature set based on an ideal charging feature set and battery charging data; The charging feature set is used as an input of a battery capacity prediction model, so that the battery capacity prediction model predicts the battery capacity of each vehicle, thereby obtaining a predicted value of the battery capacity of each vehicle; The battery capacity prediction value of each vehicle is tested based on the hypothesis testing method to obtain the battery capacity prediction result of the vehicle.
2. A battery capacity prediction method according to claim 1, characterized in that: The screening of ideal charging data based on battery charging data includes: The battery charging data that meets the following preset filtering conditions are filtered as ideal charging data: Condition 1: the vehicle needs to be discharged until the battery SOC is less than the preset first capacity threshold; Condition 2: the vehicle needs to be stationary for a preset first time before charging; Condition 3: the vehicle needs to be charged until the battery SOC is greater than the preset second capacity threshold.
3. A battery capacity prediction method according to claim 1, characterized in that: The method of obtaining an ideal charging feature set based on the ideal charging data and constructing and training a battery capacity prediction model includes: Acquire an ideal charging feature set based on ideal charging data; Calculate the ideal battery capacity of the vehicle corresponding to the ideal charging data; An initial prediction model is constructed, an ideal charging feature set is used as an input of the initial prediction model, and an ideal battery capacity is used as a label of the ideal charging feature, and then the initial prediction model is trained to obtain a battery capacity prediction model.
4. A battery capacity prediction method according to claim 3, characterized in that: The initial prediction model includes a neural network model or a random forest model.
5. A battery capacity prediction method according to claim 3, characterized in that: The step of extracting an ideal charging feature set based on ideal charging data includes: Obtaining several charging characteristics related to battery capacity in ideal charging data; The correlation between several charging characteristics and battery capacity is calculated based on correlation analysis, and an ideal charging characteristic set is extracted.
6. A battery capacity prediction method according to claim 3, characterized in that: The calculating the ideal battery capacity of the vehicle corresponding to the ideal charging data includes: Extracting the ideal charging vehicle corresponding to the ideal charging data; Calculate the ideal battery capacity C for an ideally charged vehicle n ,have: Where: C t is the SOC at the end of charging; C0 is the SOC at the start of charging; C is the rated SOC of the battery; I is the charging current.
7. A battery capacity prediction method according to claim 5, characterized in that: The extracting of the charging feature set based on the ideal charging feature set and the battery charging data includes: Determining a feature type based on an ideal charging feature set; Feature extraction is performed on the battery charging data based on the feature type to obtain a charging feature set.
8. A battery capacity prediction method according to any one of claims 3 to 7, characterized in that: The method of testing the predicted value of the battery capacity of each vehicle based on the hypothesis testing method to obtain the predicted result of the battery capacity of the vehicle includes: The predicted value of the battery capacity of each vehicle is taken as sample x, and the ideal battery capacity of the vehicle corresponding to the ideal charging data is taken as sample y. Assuming that the mean of sample x is no different from the mean of sample y, the verification value Z is calculated according to the central limit theorem, which is: Where: μ x represents the mean of sample x; μ y represents the mean of sample y; represents the variance of sample x; represents the variance of sample y; If the verification value Z is less than the preset error value, the hypothesis is accepted, and it is considered that there is no difference between the mean of sample x and the mean of sample y, and the battery capacity prediction result of the vehicle is accurate.
9. A battery capacity prediction method according to claim 8, characterized in that: The preset error value is obtained from a preset query table based on a preset single significance level.
10. A battery capacity prediction system, characterized in that: A method for predicting battery capacity according to any one of claims 1 to 9, comprising: A data acquisition module, used to acquire battery charging data of all vehicles; A data screening module, used for screening ideal charging data based on battery charging data; A feature extraction module, used to obtain an ideal charging feature set based on the ideal charging data, and to extract a charging feature set based on the ideal charging feature set and the battery charging data; A model building module, used to build and train a battery capacity prediction model based on an ideal charging feature set; A capacity prediction module is used to use the charging feature set as an input of a battery capacity prediction model so that the battery capacity prediction model predicts the battery capacity of each vehicle, thereby obtaining a predicted value of the battery capacity of each vehicle; The prediction verification module is used to verify the battery capacity prediction value of each vehicle based on the hypothesis testing method to obtain the battery capacity prediction result of the vehicle.