Battery aging prediction method and system based on data analysis

Abstract

The invention provides a battery aging prediction method and system based on data analysis, and relates to the field of battery aging prediction, and the method comprises the steps: carrying out the clustering of use features, obtaining a multivariate use feature ratio, and constructing a battery aging predictor based on the multivariate use feature ratio; performing extreme driving compensation analysis according to the real-time driving data, and outputting a driving compensation coefficient; inputting the battery monitoring data, the real-time driving data and the environment data into a battery aging predictor, and outputting a predicted battery capacity attenuation proportion; and correcting the predicted battery capacity attenuation proportion according to the driving compensation coefficient to obtain a corrected battery capacity attenuation proportion as a prediction result. The method and the device aim at solving the technical problem that a prediction result is not accurate enough due to the fact that a traditional battery aging prediction model is usually trained based on data under a conventional driving condition and extreme driving behaviors possibly occurring in actual driving are ignored, and the accuracy and the reliability of battery aging prediction under the extreme driving condition can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of battery aging prediction, and in particular to a battery aging prediction method and system based on data analysis. Background Art

[0002] The battery aging process is the result of the combined action of multiple factors, including the number of battery charge and discharge times, temperature fluctuations, charging speed, discharge depth, usage environment, and driving behavior. Battery capacity decay and internal resistance increase are the main indicators for measuring the degree of battery aging, and these decay processes are significantly affected by driving behavior. For example, frequent sudden acceleration and sudden braking will increase the instantaneous load of the battery, causing the internal temperature of the battery to rise rapidly, accelerating the decay of the battery's chemical reaction; and long-term high-speed driving will keep the battery in a high-load state, causing the electrochemical reaction inside the battery to intensify, which also accelerates battery aging.

[0003] Although there are many battery aging prediction methods based on standard driving behavior, these methods usually rely on prediction models trained based on data from conventional driving scenarios, but they fail to fully consider the actual impact of different drivers' personalized driving behaviors and different environmental factors on battery degradation. Therefore, traditional battery aging prediction models have great limitations and cannot accurately predict the battery's degradation rate and health status under extreme driving conditions. Summary of the invention

[0004] The purpose of the present invention is to provide a battery aging prediction method and system based on data analysis to solve the technical problem that the traditional battery aging prediction model is usually trained based on data under normal driving conditions, ignoring the extreme driving behaviors that may occur in actual driving, resulting in inaccurate prediction results, including: In a first aspect, the present invention provides a battery aging prediction method based on data analysis, comprising: clustering usage features according to historical usage data of a target vehicle, obtaining a proportion of multiple usage features, and constructing a battery aging predictor based on the proportion of multiple usage features; when the vehicle is driven vigorously, collecting battery monitoring data, real-time driving data, and environmental data within a preset time zone, performing extreme driving compensation analysis based on the real-time driving data, and outputting a driving compensation coefficient; inputting the battery monitoring data, real-time driving data, and environmental data into the battery aging predictor, and outputting a predicted battery capacity attenuation ratio; correcting the predicted battery capacity attenuation ratio according to the driving compensation coefficient, and obtaining a corrected battery capacity attenuation ratio as a battery aging prediction result for the preset time zone.

[0005] Preferably, the battery aging prediction method based on data analysis also includes: configuring usage characteristic indicators, wherein the usage characteristic indicators include driving surface characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics; based on the usage characteristic indicators, performing usage characteristic clustering according to historical usage data of the target vehicle to obtain multivariate usage characteristic proportions, wherein the multivariate usage characteristic proportions include road surface characteristic proportions, driving characteristic proportions, acceleration characteristic proportions, braking characteristic proportions and charging characteristic proportions.

[0006] Preferably, the battery aging prediction method based on data analysis also includes: randomly selecting a first usage characteristic indicator and obtaining multiple first characteristic intervals of the first usage characteristic indicator; performing feature extraction on historical usage data according to the first usage characteristic indicator to obtain a first usage data set; based on the first usage data set, counting the feature proportions within the multiple first characteristic intervals, constructing a first usage characteristic proportion, and adding it to the multivariate usage characteristic proportion.

[0007] Preferably, the battery aging prediction method based on data analysis also includes: configuring a feature tolerance interval, and expanding the proportion of the multivariate usage features according to the feature tolerance interval to obtain an expanded multivariate usage feature proportion; taking the expanded multivariate usage feature proportion as a conditional constraint and the target vehicle model as a body constraint, performing information retrieval based on big data to obtain a sample battery monitoring data set, a sample driving data set and a sample environmental data set, and obtaining the battery capacity attenuation ratio of different sample battery monitoring data, sample driving data and sample environmental data in a historical time zone, and constructing a sample battery capacity attenuation ratio set; using the sample battery monitoring data set, sample driving data set, sample environmental data set and sample battery capacity attenuation ratio set to supervise the feedforward neural network until the model converges to obtain the battery aging predictor.

[0008] Preferably, the battery aging prediction method based on data analysis also includes: using the sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set as training data, and dividing them into K parts, to obtain K training sets, wherein K is an integer greater than 10; using the K training sets, respectively performing supervised training and verification on the feedforward neural network until the model converges, to obtain K battery aging prediction branches, and integrating to construct the battery aging predictor.

[0009] Preferably, the battery aging prediction method based on data analysis also includes: performing driving characteristics, acceleration characteristics and braking characteristics analysis according to the real-time driving data, and calculating the average driving speed, the average acceleration and the average braking speed; calculating the historical average driving speed, the historical average acceleration and the historical average braking speed according to the historical usage data of the target vehicle; and calculating the driving compensation coefficient according to the average driving speed, the average acceleration, the average braking speed, the historical average driving speed, the historical average acceleration and the historical average braking speed.

[0010] Preferably, the battery aging prediction method based on data analysis also includes: setting the ratio of the average driving speed to the average historical driving speed as a driving compensation coefficient, setting the ratio of the average acceleration to the average historical acceleration as an acceleration compensation coefficient, and setting the ratio of the average braking speed to the average historical braking speed as a braking compensation coefficient; and obtaining the driving compensation coefficient by weighted calculation based on the driving compensation coefficient, the acceleration compensation coefficient and the braking compensation coefficient.

[0011] Preferably, the battery aging prediction method based on data analysis also includes: setting the ratio of the driving compensation coefficient to the historical maximum driving compensation coefficient as a compensation scale coefficient, and multiplying the compensation scale coefficient by K and rounding it to obtain Q, where Q is an integer greater than or equal to 1 and less than or equal to K; randomly selecting Q battery aging prediction branches from the K battery aging prediction branches of the battery aging predictor, performing prediction according to the battery monitoring data, real-time driving data and environmental data, outputting Q predicted attenuation ratios, and obtaining the predicted battery capacity attenuation ratio after mean calculation.

[0012] Preferably, the battery aging prediction method based on data analysis further includes: taking the product of the driving compensation coefficient and the predicted battery capacity attenuation ratio as the corrected battery capacity attenuation ratio.

[0013] In a second aspect, the present invention further provides a battery aging prediction system based on data analysis, which is used to execute a battery aging prediction method based on data analysis as described in the first aspect, including: a battery aging predictor construction module, which is used to cluster usage features according to historical usage data of the target vehicle, obtain the proportion of multivariate usage features, and construct a battery aging predictor based on the proportion of multivariate usage features; an extreme driving compensation analysis module, which is used to collect battery monitoring data, real-time driving data and environmental data within a preset time zone when the vehicle is driving vigorously, perform extreme driving compensation analysis based on the real-time driving data, and output a driving compensation coefficient; a battery capacity attenuation prediction module, which is used to input the battery monitoring data, real-time driving data and environmental data into the battery aging predictor, and output a predicted battery capacity attenuation ratio; a battery capacity attenuation ratio correction module, which is used to correct the predicted battery capacity attenuation ratio according to the driving compensation coefficient to obtain a corrected battery capacity attenuation ratio as the battery aging prediction result for the preset time zone.

[0014] The embodiments of the present invention include the following advantages: By clustering usage features according to the historical usage data of the target vehicle, the proportion of multiple usage features is obtained, and a battery aging predictor is constructed based on the proportion of multiple usage features; then, when the vehicle is driven vigorously, the battery monitoring data, real-time driving data and environmental data in the preset time zone are collected, and extreme driving compensation analysis is performed according to the real-time driving data, and a driving compensation coefficient is output; then the battery monitoring data, real-time driving data and environmental data are input into the battery aging predictor, and a predicted battery capacity attenuation ratio is output; finally, the predicted battery capacity attenuation ratio is corrected according to the driving compensation coefficient, and the corrected battery capacity attenuation ratio is obtained as the battery aging prediction result of the preset time zone. In other words, by introducing the driving behavior compensation coefficient to compensate for the battery aging prediction result under intense driving conditions, the accuracy and reliability of battery aging prediction under extreme driving conditions can be effectively improved, thereby improving the overall performance and reliability of electric vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of the steps of a battery aging prediction method based on data analysis of the present invention; Figure 2 The present invention is a structural schematic diagram of a battery aging prediction system based on data analysis.

[0016] Description of reference numerals: A battery aging predictor construction module 11, an extreme driving compensation analysis module 12, a battery capacity attenuation prediction module 13, and a battery capacity attenuation ratio correction module 14. DETAILED DESCRIPTION

[0017] The present invention provides a battery aging prediction method and system based on data analysis, which solves the technical problem that the traditional battery aging prediction model is usually trained based on data under normal driving conditions, ignoring the extreme driving behaviors that may occur in actual driving, resulting in inaccurate prediction results. By introducing a driving behavior compensation coefficient to compensate for the battery aging prediction results under intense driving conditions, the accuracy and reliability of battery aging prediction under extreme driving conditions can be effectively improved, thereby improving the overall performance and reliability of electric vehicles.

[0018] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. 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. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0019] For example, please refer to the attached Figure 1 The present invention provides a battery aging prediction method based on data analysis, which is applied to a battery aging prediction system based on data analysis, and specifically includes the following steps: S1: Clustering usage features according to historical usage data of the target vehicle to obtain multivariate usage feature proportions, and constructing a battery aging predictor based on the multivariate usage feature proportions.

[0020] Further, step S1 of the present invention further comprises: S11: configuring usage characteristic indicators, wherein the usage characteristic indicators include driving road characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics.

[0021] Specifically, in the prediction of battery aging, the configuration of the use of characteristic indicators is very critical. These characteristics can fully understand the use of the vehicle, especially the impact of driver behavior and environmental conditions on battery aging. Reasonable configuration of the use of characteristic indicators can not only improve the prediction accuracy, but also realize personalized adjustment and optimize battery management. First, configure the use of characteristic indicators, where the use of characteristic indicators includes driving surface characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics. Driving surface characteristics include road surface type, driving characteristics include driving speed, acceleration characteristics include acceleration force and acceleration frequency, braking characteristics include braking frequency and braking intensity, and charging characteristics include charging frequency and charging depth. By configuring driving surface characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics, the key factors affecting battery aging can be fully grasped; these characteristics can not only help predict the battery degradation trend, but also be dynamically adjusted according to real-time data to achieve more accurate battery management and aging prediction.

[0022] S12: Based on the usage characteristic index, the usage characteristics are clustered according to the historical usage data of the target vehicle to obtain the multi-dimensional usage characteristic proportion, wherein the multi-dimensional usage characteristic proportion includes the road surface characteristic proportion, the driving characteristic proportion, the acceleration characteristic proportion, the braking characteristic proportion and the charging characteristic proportion.

[0023] Further, step S12 of the present invention further includes: S121: Randomly select a first usage characteristic indicator and obtain multiple first characteristic intervals of the first usage characteristic indicator; S122: Perform feature extraction on historical usage data according to the first usage characteristic indicator to obtain a first usage data set; S123: Based on the first usage data set, count the feature proportions in the multiple first characteristic intervals, construct a first usage characteristic proportion, and add it to the multivariate usage characteristic proportion.

[0024] Specifically, any one of the multiple usage characteristic indicators is randomly selected as the first usage characteristic indicator. This indicator can be one of the driving road characteristics, driving characteristics, acceleration characteristics, braking characteristics or charging characteristics. Taking driving speed as an example, this is the selected first usage characteristic indicator. In order to better analyze the first usage characteristic indicator (such as driving speed), the indicator can be divided into multiple characteristic intervals. These characteristic intervals will help to understand the load and degradation effects of the battery at different driving speeds, for example, a low speed interval (0 to 30 km / h), a medium speed interval (31 to 60 km / h), a high speed interval (61 to 120 km / h), and an ultra-high speed interval (greater than 120 km / h).

[0025] Then, the first usage data set is extracted based on the historical usage data. This means that the system will perform data screening and statistics on the characteristic indicator of driving speed from the historical records of the vehicle, and extract relevant values. The historical usage data may include information such as the real-time driving speed, driving duration, and driving route of the vehicle to obtain the first usage data set. Further based on the first usage data set, the feature proportions within the multiple first feature intervals are counted, that is, the proportion of each feature interval in the overall data is counted; based on the statistical feature proportions, a first usage feature proportion is constructed, which reflects the usage of the vehicle in different usage scenarios, where the driving feature proportion table is shown in Table 1: Table 1: Driving characteristics ratio

[0026] This table intuitively lists the proportion of each speed range, which is convenient for subsequent analysis and application.

[0027] Using the same method as that for obtaining the first usage feature proportion (driving feature proportion), feature cluster analysis is performed on other usage feature indicators (driving road surface characteristics, acceleration characteristics, braking characteristics and charging characteristics) in turn to obtain the multivariate usage feature proportion, where the multivariate usage feature proportion includes the road surface characteristic proportion, the driving feature proportion, the acceleration characteristic proportion, the braking characteristic proportion and the charging characteristic proportion. For example, for the driving road surface characteristics, feature extraction is performed according to different road surface types (such as urban roads, highways, mountain roads, etc.), and the proportion of each road surface type is counted; for the acceleration characteristics, intervals are divided according to different acceleration intensities or frequencies, and can usually be classified by sudden acceleration, smooth acceleration and slow acceleration; for the braking characteristics, the characteristics are distinguished according to the intensity of braking (sudden braking, smooth braking and light braking); for the charging characteristics, the intervals can be divided according to the charging method (such as fast charging, slow charging and charging frequency), where the multivariate usage feature proportion table is shown in Table 2: Table 2: Percentage of multiple usage characteristics

[0028] Further, step S1 of the present invention further comprises: S13: configuring a feature tolerance interval, and expanding the proportion of the multi-use features according to the feature tolerance interval to obtain an expanded proportion of the multi-use features; S14: taking the expanded proportion of the multi-use features as a conditional constraint and the target vehicle model as a body constraint, performing information retrieval based on big data, obtaining a sample battery monitoring data set, a sample driving data set and a sample environmental data set, as well as obtaining the battery capacity attenuation ratio of different sample battery monitoring data, sample driving data and sample environmental data in a historical time zone, and constructing a sample battery capacity attenuation ratio set.

[0029] Specifically, a feature tolerance interval is configured. The feature tolerance interval refers to a certain range of variation allowed within the interval of each usage feature to reflect fluctuations or abnormal situations that may occur during actual driving. Then, the proportion of the multiple usage features is expanded according to the feature tolerance interval, that is, based on the results of the tolerance interval, the proportion of each feature interval is recalculated to ensure the accuracy and breadth of the data. This method can more comprehensively capture the characteristics of the vehicle under different usage conditions. For example, the driving speed interval is expanded. Assuming that the original interval is 0 to 30 km / h and the tolerance interval is ±5 km / h, the expanded interval becomes 25 to 35 km / h. Similarly, for other acceleration, braking and other features, similar tolerance extensions are performed to obtain new feature interval proportions.

[0030] Next, taking the proportion of the extended multivariate usage features as a conditional constraint and the target vehicle model as a body constraint, information retrieval is performed based on big data to obtain sample data similar to the target vehicle, and a sample battery monitoring data set, a sample driving data set, and a sample environment data set are obtained, wherein the sample battery monitoring data includes battery voltage, current, temperature, SOC (state of charge) and other data; the sample driving data includes driver speed, acceleration, braking, driving route and other data; the sample environment data includes temperature, humidity, terrain and other external environment data; further obtain the battery capacity attenuation ratio of different sample battery monitoring data, sample driving data and sample environment data in the historical time zone, and construct a sample battery capacity attenuation ratio set.

[0031] By configuring and expanding the feature tolerance interval, the multi-use characteristics of the vehicle can be described more accurately. Combining the target vehicle model as a body constraint, information retrieval based on big data is performed to finally obtain sample battery monitoring data sets, sample driving data sets, and sample environmental data sets, and calculate the capacity attenuation ratio of different sample batteries to form a complete sample attenuation ratio set. This process can greatly improve the accuracy and reliability of battery aging prediction.

[0032] S15: Using the sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set, supervised training is performed on the feedforward neural network until the model converges to obtain the battery aging predictor.

[0033] Further, step S15 of the present invention further includes: S151: Use the sample battery monitoring data set, sample driving data set, sample environment data set and sample battery capacity attenuation ratio set as training data, and divide them into K equal parts to obtain K training sets, where K is an integer greater than 10; S152: Use the K training sets to perform supervised training and verification on the feedforward neural network respectively until the model converges, and obtain K battery aging prediction branches, which are integrated to construct the battery aging predictor.

[0034] Specifically, the sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set are used as training data, and the training data is divided into K equal parts, where K is an integer greater than 10, and the specific value of K can be set according to actual needs, for example, K is set to 20; K training sets are obtained.

[0035] A feedforward neural network is a common artificial neural network structure, which usually consists of an input layer, one or more hidden layers, and an output layer. In this model, we will use the input layer to receive sample data (such as battery monitoring data, driving data, and environmental data), calculate through the hidden layer, and finally predict the capacity attenuation ratio of the battery through the output layer. Then, the K training sets are used to supervise the training and verification of the feedforward neural network. In each training process, the training set is used to train the model and optimize the weight parameters of the neural network. At the same time, the verification set is used to verify the performance of the model to ensure that the model has good prediction ability on unseen data. During training, gradient descent or other optimization algorithms (such as Adam optimizer) are used to update the weights of the neural network. After each training, the verification set is used to evaluate the performance of the model and calculate the error. The training process will continue until the performance of the model on the verification set no longer improves, indicating that the model has converged. Then, through K-fold cross-validation training, K different model branches are obtained. Each model branch is trained based on different training data and verified on different verification sets. Finally, a battery aging predictor is constructed based on the combination of K battery aging prediction branches.

[0036] By integrating K training branches to build a battery aging predictor, more robust and accurate battery aging prediction results can be obtained, providing strong support for battery management of electric vehicles.

[0037] S2: When the vehicle is driven vigorously, the battery monitoring data, real-time driving data and environmental data in a preset time zone are collected, and an extreme driving compensation analysis is performed based on the real-time driving data to output a driving compensation coefficient.

[0038] Further, step S2 of the present invention also includes: S21: Analyze the driving characteristics, acceleration characteristics and braking characteristics according to the real-time driving data, and calculate the average driving speed, average acceleration and average braking speed; S22: Calculate the historical driving speed average, historical acceleration average and historical braking speed average according to the historical usage data of the target vehicle.

[0039] Specifically, when the vehicle is driven vigorously, the battery monitoring data, real-time driving data and environmental data within the preset time zone (such as the last hour) are collected. The battery monitoring data includes battery voltage, battery current, battery temperature and SOC (state of charge), etc. The real-time driving data includes vehicle speed, driving characteristics, braking characteristics, etc.; the environmental data includes temperature, humidity, etc. By collecting the battery monitoring data, real-time driving data and environmental data within the preset time zone, we can fully understand the performance changes of the battery under extreme driving conditions. These data can not only provide accurate input data for battery aging prediction, but also help predict the battery's decay rate and remaining service life under different environments and driving behaviors.

[0040] Next, the driving characteristics, acceleration characteristics and braking characteristics are analyzed based on the real-time driving data, and the average driving speed, average acceleration and average braking speed are calculated. The average driving speed represents the average driving speed of the vehicle over a period of time; the average acceleration represents the average acceleration of the vehicle over a period of time, reflecting the acceleration intensity of the vehicle; the average braking speed represents the average braking force of the vehicle over a period of time. This indicator reflects the frequency and intensity of braking. On the other hand, based on the historical usage data of the target vehicle, the historical average driving speed, historical average acceleration and historical average braking speed are calculated. The historical average driving speed represents the average driving speed of the vehicle over a historical period of time; the historical average acceleration represents the average acceleration of the vehicle over a historical period of time; the historical average braking speed represents the average braking of the vehicle over a historical period of time.

[0041] S23: Calculate the driving compensation coefficient according to the average driving speed, the average acceleration, the average braking speed, the historical average driving speed, the historical average acceleration and the historical average braking speed.

[0042] Further, step S23 of the present invention further includes: S231: The ratio of the average driving speed to the average historical driving speed is set as the driving compensation coefficient, the ratio of the average acceleration to the average historical acceleration is set as the acceleration compensation coefficient, and the ratio of the average braking speed to the average historical braking speed is set as the braking compensation coefficient; S232: The driving compensation coefficient is obtained by weighted calculation based on the driving compensation coefficient, the acceleration compensation coefficient and the braking compensation coefficient.

[0043] Specifically, first, the ratio of the average driving speed to the average historical driving speed is set as the driving compensation coefficient, the ratio of the average acceleration to the average historical acceleration is set as the acceleration compensation coefficient, and the ratio of the average braking speed to the average historical braking speed is set as the braking compensation coefficient. The driving compensation coefficient reflects the deviation between the current driving speed of the vehicle and its historical driving state; the acceleration compensation coefficient reflects the difference between the current acceleration intensity and the historical acceleration state; and the braking compensation coefficient reflects the difference between the current braking force and the historical braking conditions.

[0044] Next, configure the driving compensation weight, acceleration compensation weight and braking compensation weight, where the sum of the driving compensation weight, acceleration compensation weight and braking compensation weight is 1, reflecting the degree of influence of each feature on battery aging. These weights can be adjusted according to actual experience or experimental data to ensure that the contribution of each feature to the compensation coefficient is reasonable. Then, according to the driving compensation weight, acceleration compensation weight and braking compensation weight, perform weighted calculation on the driving compensation coefficient, acceleration compensation coefficient and braking compensation coefficient, and use the weighted calculation result as the driving compensation coefficient.

[0045] S3: Input the battery monitoring data, real-time driving data and environmental data into the battery aging predictor, and output a predicted battery capacity attenuation ratio.

[0046] Further, step S3 of the present invention also includes: S31: setting the ratio of the driving compensation coefficient to the historical maximum driving compensation coefficient as the compensation scale coefficient, and multiplying the compensation scale coefficient by K and rounding it to obtain Q, where Q is an integer greater than or equal to 1 and less than or equal to K; S32: randomly selecting Q battery aging prediction branches from the K battery aging prediction branches of the battery aging predictor, performing predictions based on the battery monitoring data, real-time driving data and environmental data, outputting Q predicted attenuation ratios, and obtaining the predicted battery capacity attenuation ratio after mean calculation.

[0047] Specifically, a driving compensation coefficient and a historical maximum driving compensation coefficient are obtained, wherein the driving compensation coefficient is a coefficient calculated based on driving, acceleration and braking behaviors, and the historical maximum driving compensation coefficient is the value of the maximum driving compensation coefficient in all historical data; then, the ratio of the driving compensation coefficient to the historical maximum driving compensation coefficient is set as the compensation scale coefficient, and the compensation scale coefficient is multiplied by K and rounded to obtain Q, where Q is an integer greater than or equal to 1 and less than or equal to K. For example, assuming that the driving compensation coefficient is 1.5, the historical maximum driving compensation coefficient is 3, and K is 20, then Q is 1.5 / 3*20, which is equal to 10.

[0048] Then, Q battery aging prediction branches are randomly selected from the K battery aging prediction branches of the battery aging predictor, and the battery monitoring data, real-time driving data and environmental data are respectively input into the Q battery aging prediction branches for prediction, Q predicted attenuation ratios are output, and the output Q predicted attenuation ratios are averaged to obtain the predicted battery capacity attenuation ratio.

[0049] The compensation scale coefficient is calculated according to the current driving state, and an adapted number of battery aging prediction branches are selected according to the compensation scale coefficient for prediction analysis. That is, when the compensation scale coefficient is large, it means that the gap between the current driving conditions and the historical maximum driving conditions is large. At this time, more prediction branches can be selected to capture more battery aging information and ensure the accuracy of the prediction results; when the compensation scale coefficient is small, it means that the current driving conditions are close to the historical conditions, and the number of required prediction branches can be reduced, thereby saving computing resources; the number of selected prediction branches is dynamically adjusted by the compensation scale coefficient, which can reduce unnecessary calculations and save computing resources while ensuring the accuracy of battery aging prediction.

[0050] S4: Correcting the predicted battery capacity attenuation ratio according to the driving compensation coefficient to obtain a corrected battery capacity attenuation ratio as the battery aging prediction result in the preset time zone.

[0051] Further, step S4 of the present invention also includes: S41: taking the product of the driving compensation coefficient and the predicted battery capacity attenuation ratio as the corrected battery capacity attenuation ratio.

[0052] Specifically, the predicted battery capacity attenuation ratio is corrected according to the driving compensation coefficient, that is, the driving compensation coefficient is multiplied by the predicted battery capacity attenuation ratio to obtain the corrected battery capacity attenuation ratio. This correction process makes the prediction result more consistent with the influence of the current driving behavior, especially under extreme driving behaviors (such as fierce acceleration, high-speed driving, frequent braking, etc.), it can more accurately reflect the actual aging of the battery. Finally, the corrected battery capacity attenuation ratio is used as the battery aging prediction result for the preset time zone. By multiplying the driving compensation coefficient by the predicted battery capacity attenuation ratio, a more accurate battery aging prediction result is obtained. This process optimizes the battery attenuation prediction through dynamic compensation for driving behavior, making it more suitable for intense driving conditions and improving the reliability and accuracy of the battery management system.

[0053] In summary, the battery aging prediction method based on data analysis provided by the present invention has the following technical effects: By clustering usage features according to the historical usage data of the target vehicle, the proportion of multiple usage features is obtained, and a battery aging predictor is constructed based on the proportion of multiple usage features; then, when the vehicle is driven vigorously, the battery monitoring data, real-time driving data and environmental data in the preset time zone are collected, and extreme driving compensation analysis is performed according to the real-time driving data, and a driving compensation coefficient is output; then the battery monitoring data, real-time driving data and environmental data are input into the battery aging predictor, and a predicted battery capacity attenuation ratio is output; finally, the predicted battery capacity attenuation ratio is corrected according to the driving compensation coefficient, and the corrected battery capacity attenuation ratio is obtained as the battery aging prediction result of the preset time zone. In other words, by introducing the driving behavior compensation coefficient to compensate for the battery aging prediction result under intense driving conditions, the accuracy and reliability of battery aging prediction under extreme driving conditions can be effectively improved, thereby improving the overall performance and reliability of electric vehicles.

[0054] Embodiment 2: Based on the same inventive concept as the battery aging prediction method based on data analysis in the above embodiment, the present invention also provides a battery aging prediction system based on data analysis, please refer to the attached Figure 2 , including: a battery aging predictor construction module 11, used to cluster usage features according to the historical usage data of the target vehicle, obtain the proportion of multiple usage features, and construct a battery aging predictor based on the proportion of multiple usage features; an extreme driving compensation analysis module 12, used to collect battery monitoring data, real-time driving data and environmental data in a preset time zone when the vehicle is driving vigorously, perform extreme driving compensation analysis according to the real-time driving data, and output a driving compensation coefficient; a battery capacity decay prediction module 13, used to input the battery monitoring data, real-time driving data and environmental data into the battery aging predictor, and output a predicted battery capacity decay ratio; a battery capacity decay ratio correction module 14, used to correct the predicted battery capacity decay ratio according to the driving compensation coefficient, and obtain a corrected battery capacity decay ratio as the battery aging prediction result of the preset time zone.

[0055] Furthermore, the battery aging prediction system based on data analysis is also used to: configure usage characteristic indicators, wherein the usage characteristic indicators include driving surface characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics; based on the usage characteristic indicators, perform usage characteristic clustering according to historical usage data of the target vehicle to obtain multivariate usage characteristic proportions, wherein the multivariate usage characteristic proportions include road surface characteristic proportions, driving characteristic proportions, acceleration characteristic proportions, braking characteristic proportions and charging characteristic proportions.

[0056] Furthermore, the battery aging prediction system based on data analysis is also used to: randomly select a first usage characteristic indicator and obtain multiple first characteristic intervals of the first usage characteristic indicator; perform feature extraction on historical usage data according to the first usage characteristic indicator to obtain a first usage data set; based on the first usage data set, count the feature proportions within the multiple first characteristic intervals, construct a first usage characteristic proportion, and add it to the multivariate usage characteristic proportion.

[0057] Furthermore, the battery aging prediction system based on data analysis is also used to: configure a feature tolerance interval, and expand the proportion of the multivariate usage features according to the feature tolerance interval to obtain an expanded multivariate usage feature proportion; use the expanded multivariate usage feature proportion as a conditional constraint and the target vehicle model as a body constraint, perform information retrieval based on big data, obtain a sample battery monitoring data set, a sample driving data set and a sample environmental data set, and obtain the battery capacity attenuation ratio of different sample battery monitoring data, sample driving data and sample environmental data in a historical time zone, and construct a sample battery capacity attenuation ratio set; use the sample battery monitoring data set, sample driving data set, sample environmental data set and sample battery capacity attenuation ratio set to supervise the feedforward neural network until the model converges to obtain the battery aging predictor.

[0058] Furthermore, the battery aging prediction system based on data analysis is also used to: use the sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set as training data, and divide them into K equal parts to obtain K training sets, wherein K is an integer greater than 10; use the K training sets to perform supervised training and verification on the feedforward neural network respectively until the model converges, to obtain K battery aging prediction branches, and integrate to construct the battery aging predictor.

[0059] Furthermore, the battery aging prediction system based on data analysis is also used to: perform driving characteristics, acceleration characteristics and braking characteristics analysis according to the real-time driving data, and calculate the average driving speed, the average acceleration and the average braking speed; calculate the historical average driving speed, the historical average acceleration and the historical average braking speed according to the historical usage data of the target vehicle; calculate the driving compensation coefficient according to the average driving speed, the average acceleration, the average braking speed, the historical average driving speed, the historical average acceleration and the historical average braking speed.

[0060] Furthermore, the battery aging prediction system based on data analysis is also used to: set the ratio of the average driving speed to the average historical driving speed as a driving compensation coefficient, set the ratio of the average acceleration to the average historical acceleration as an acceleration compensation coefficient, and set the ratio of the average braking speed to the average historical braking speed as a braking compensation coefficient; and obtain the driving compensation coefficient by weighted calculation based on the driving compensation coefficient, the acceleration compensation coefficient and the braking compensation coefficient.

[0061] Furthermore, the battery aging prediction system based on data analysis is also used to: set the ratio of the driving compensation coefficient to the historical maximum driving compensation coefficient as a compensation scale coefficient, and multiply the compensation scale coefficient by K and round it to obtain Q, where Q is an integer greater than or equal to 1 and less than or equal to K; randomly select Q battery aging prediction branches from the K battery aging prediction branches of the battery aging predictor, perform predictions based on the battery monitoring data, real-time driving data and environmental data, output Q predicted attenuation ratios, and obtain the predicted battery capacity attenuation ratio after mean calculation.

[0062] Furthermore, the battery aging prediction system based on data analysis is also used to: use the product of the driving compensation coefficient and the predicted battery capacity attenuation ratio as the corrected battery capacity attenuation ratio.

[0063] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The battery aging prediction method based on data analysis and the specific examples in the aforementioned embodiment 1 are also applicable to a battery aging prediction system based on data analysis in this embodiment. Through the aforementioned detailed description of a battery aging prediction method based on data analysis, those skilled in the art can clearly know a battery aging prediction system based on data analysis in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0064] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0065] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. A battery aging prediction method based on data analysis, characterized in that the method include: Clustering usage features according to historical usage data of the target vehicle to obtain multivariate usage feature proportions, and building a battery aging predictor based on the multivariate usage feature proportions; When the vehicle is driven vigorously, the battery monitoring data, real-time driving data and environmental data in a preset time zone are collected, and an extreme driving compensation analysis is performed based on the real-time driving data to output a driving compensation coefficient; Inputting the battery monitoring data, real-time driving data and environmental data into the battery aging predictor, and outputting a predicted battery capacity attenuation ratio; The predicted battery capacity attenuation ratio is corrected according to the driving compensation coefficient to obtain a corrected battery capacity attenuation ratio as the battery aging prediction result in the preset time zone.

2. The battery aging prediction method based on data analysis according to claim 1, characterized in that: Use the historical usage data of the target vehicle to cluster usage features and obtain the proportion of multiple usage features, including: Configure usage characteristic indicators, where the usage characteristic indicators include driving road characteristics, driving characteristics, acceleration characteristics, braking characteristics and charging characteristics; Based on the usage characteristic index, usage characteristic clustering is performed according to the historical usage data of the target vehicle to obtain the proportion of multiple usage characteristics, wherein the proportion of multiple usage characteristics includes the proportion of road surface characteristics, the proportion of driving characteristics, the proportion of acceleration characteristics, the proportion of braking characteristics and the proportion of charging characteristics.

3. The battery aging prediction method based on data analysis according to claim 2, characterized in that: Get the proportion of multiple usage features, including: Randomly selecting a first usage characteristic indicator, and obtaining a plurality of first characteristic intervals of the first usage characteristic indicator; Extracting features from the historical usage data according to the first usage feature indicator to obtain a first usage data set; Based on the first usage data set, feature proportions within the plurality of first feature intervals are counted to construct a first usage feature proportion, which is added to the multivariate usage feature proportion.

4. The battery aging prediction method based on data analysis according to claim 1, characterized in that: A battery aging predictor is constructed based on the multivariate usage feature proportions, including: configuring a feature tolerance interval, and expanding the proportion of the multiple usage features according to the feature tolerance interval to obtain an expanded proportion of the multiple usage features; Taking the extended multivariate usage feature proportion as a conditional constraint and the target vehicle model as a vehicle body constraint, information retrieval is performed based on big data to obtain a sample battery monitoring data set, a sample driving data set, and a sample environment data set, as well as the battery capacity attenuation ratio of different sample battery monitoring data, sample driving data, and sample environment data in a historical time zone, to construct a sample battery capacity attenuation ratio set; The sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set are used to perform supervised training on a feedforward neural network until the model converges, thereby obtaining the battery aging predictor.

5. The battery aging prediction method based on data analysis according to claim 4 is characterized in that: Obtaining the battery aging predictor, comprising: The sample battery monitoring data set, the sample driving data set, the sample environment data set and the sample battery capacity attenuation ratio set are used as training data, and are equally divided into K parts to obtain K training sets, where K is an integer greater than 10; The K training sets are used to perform supervised training and verification on the feedforward neural network respectively until the model converges, thereby obtaining K battery aging prediction branches, which are integrated to construct the battery aging predictor.

6. The battery aging prediction method based on data analysis according to claim 5, characterized in that: Perform extreme driving compensation analysis according to the real-time driving data and output a driving compensation coefficient, including: Analyzing driving characteristics, acceleration characteristics, and braking characteristics according to the real-time driving data, and calculating the average driving speed, the average acceleration, and the average braking speed; According to the historical usage data of the target vehicle, the historical average driving speed, historical average acceleration and historical average braking speed are calculated; The driving compensation coefficient is calculated based on the average driving speed, the average acceleration, the average braking speed, the historical average driving speed, the historical average acceleration and the historical average braking speed.

7. The battery aging prediction method based on data analysis according to claim 6, characterized in that: The driving compensation coefficient is calculated, including: The ratio of the driving speed average to the historical driving speed average is set as the driving compensation coefficient, the ratio of the acceleration average to the historical acceleration average is set as the acceleration compensation coefficient, and the ratio of the braking speed average to the historical braking speed average is set as the braking compensation coefficient; The driving compensation coefficient is obtained by weighted calculation according to the travel compensation coefficient, the acceleration compensation coefficient and the braking compensation coefficient.

8. The battery aging prediction method based on data analysis according to claim 7, characterized in that: The battery monitoring data, real-time driving data and environmental data are input into the battery aging predictor, and the predicted battery capacity attenuation ratio is output, including: The ratio of the driving compensation coefficient to the historical maximum driving compensation coefficient is set as the compensation scale coefficient, and the compensation scale coefficient is multiplied by K and rounded to obtain Q, where Q is an integer greater than or equal to 1 and less than or equal to K; Q battery aging prediction branches are randomly selected from the K battery aging prediction branches of the battery aging predictor, prediction is performed according to the battery monitoring data, real-time driving data and environmental data, Q predicted attenuation ratios are output, and the predicted battery capacity attenuation ratio is obtained after mean calculation.

9. The battery aging prediction method based on data analysis according to claim 8, characterized in that: The product of the driving compensation coefficient and the predicted battery capacity attenuation ratio is used as the corrected battery capacity attenuation ratio.

10. The battery aging prediction system based on data analysis is characterized by: The steps for implementing the battery aging prediction method based on data analysis as described in any one of claims 1 to 9 include: A battery aging predictor building module, used to cluster usage features according to historical usage data of the target vehicle, obtain multivariate usage feature proportions, and build a battery aging predictor based on the multivariate usage feature proportions; An extreme driving compensation analysis module is used to collect battery monitoring data, real-time driving data and environmental data within a preset time zone when the vehicle is driven vigorously, perform extreme driving compensation analysis based on the real-time driving data, and output a driving compensation coefficient; A battery capacity attenuation prediction module, used to input the battery monitoring data, real-time driving data and environmental data into the battery aging predictor, and output a predicted battery capacity attenuation ratio; The battery capacity attenuation ratio correction module is used to correct the predicted battery capacity attenuation ratio according to the driving compensation coefficient to obtain a corrected battery capacity attenuation ratio as the battery aging prediction result in the preset time zone.

Images