Battery state detection method and device

Abstract

The embodiment of the invention provides a battery state detection method and device, and the method comprises the steps: obtaining historical charging data of a vehicle battery in a historical time period; data analysis is conducted on the historical charging data, multiple charging data fragments are intercepted from the historical charging data, and the multiple charging data fragments are used for representing data fragments for continuous charging of the vehicle battery within a preset charging current range; feature extraction is conducted on the multiple charging data fragments, charging feature data of the vehicle battery are obtained, and the charging feature data are used for reflecting the electric quantity change condition and the temperature change condition of the vehicle battery when charging is conducted on the target vehicle within the historical time period; the battery state detection model is used for processing the charging characteristic data, the battery state of the vehicle battery is obtained, and the battery state is used for representing whether the vehicle battery has the power shortage risk or not. The technical problem of low precision of predicting the health state of the vehicle battery in the prior art is solved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of battery health assessment, and in particular to a battery status detection method and device. Background Art

[0002] The new energy vehicle industry is currently transitioning from single-battery performance monitoring to a comprehensive vehicle health management system. As vehicles age, their capacity and health gradually deteriorate, posing a potential threat to vehicle performance and user safety.

[0003] In recent years, with the rapid development of Internet of Vehicles (IoV) technology, new energy vehicles can upload vehicle status and battery data to the cloud in real time, which makes it possible to achieve more accurate battery health predictions. However, current vehicle data collection and analysis methods often fail to fully consider the actual performance of batteries under different operating conditions, resulting in low accuracy of battery health predictions. Summary of the Invention

[0004] The embodiments of the present application provide a battery status detection method and device, aiming to improve the technical problem of low accuracy in predicting the health status of vehicle batteries in the prior art.

[0005] According to one embodiment of the present application, a battery status detection method is provided, including: obtaining historical charging data of a vehicle battery in a historical time period, wherein the historical charging data is used to characterize data generated when a power supply battery supplies power to a vehicle battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery; performing data analysis on the historical charging data, and intercepting multiple charging data segments from the historical charging data, wherein the multiple charging data segments are used to characterize data segments of continuous charging of the vehicle battery within a preset charging current range; performing feature extraction on the multiple charging data segments to obtain charging characteristic data of the vehicle battery, wherein the charging characteristic data is used to reflect the changes in the charge level and temperature of the vehicle battery when the target vehicle is charged within the historical time period; and processing the charging characteristic data using a battery status detection model to obtain a battery status of the vehicle battery, wherein the battery status is used to characterize whether the vehicle battery has a risk of power outage.

[0006] The above optional embodiments of the present application can achieve the following beneficial effects: First, by obtaining detailed historical charging data of the vehicle battery in the historical time period, we can capture the dynamic characteristics of the power supply battery when charging the vehicle battery. Since the voltage of the power supply battery is higher than that of the vehicle battery, this charging process can be carried out relatively stably and continuously, providing ideal conditions for data collection and ensuring the basis of data quality. Secondly, in-depth data analysis is performed on the historical charging data, and continuous charging data fragments within the preset charging current range are accurately intercepted from the massive data through intelligent algorithms. These fragmented data focus on specific current conditions and exclude irrelevant or low-quality data, thereby improving the pertinence and efficiency of subsequent feature extraction. Selecting data fragments within the preset charging current range can ensure that the charging process is carried out under controllable and comparable conditions, reducing interference caused by other working conditions. Next, feature extraction is performed on the multiple captured charging data segments to obtain charging feature data that can reflect the changes in the vehicle battery power and temperature. This process not only reveals the dynamic changes in battery power over time, but also considers the impact of temperature on battery performance, providing comprehensive and accurate input information for model training. The extraction of feature data, especially the focus on the two key indicators of power change and temperature change, can more accurately characterize the battery's health state and provide solid data support for the model's predictions. Finally, the charging feature data is processed using the battery status detection model to accurately determine the current status of the vehicle battery, including an assessment of whether there is a risk of low power.

[0007] Furthermore, feature extraction is performed on multiple power-recharging data segments to obtain power-recharging characteristic data of the vehicle battery, including: feature extraction is performed on the power-recharging data segments to obtain the power-recharging time period and power-recharging parameters corresponding to the power-recharging data segments; based on the power-recharging parameters, the segment charging power and segment average temperature corresponding to the power-recharging data segments are determined, wherein the segment charging power is used to characterize the charging power of the vehicle battery in the corresponding power-recharging time period, and the segment average temperature is used to characterize the average temperature of the vehicle battery in the corresponding power-recharging time period; the segment charging power is processed based on the target charging power, the power-recharging time period and the segment average temperature to obtain the charging power change of the vehicle battery in different power-recharging time periods, wherein the target charging power is used to characterize the segment charging power corresponding to the first power-recharging data segment among the multiple power-recharging data segments; based on the charging power change and the target average temperature, the power-recharging characteristic data is constructed, wherein the target average temperature is used to characterize the segment average temperature corresponding to the first power-recharging data segment among the multiple power-recharging data segments.

[0008] The above optional embodiments of the present application can achieve the following beneficial effects: by performing in-depth feature extraction on the charging data segments, not only the specific time period of charging is obtained, but also the key parameters of the charging process are accurately captured. This refined data processing method can comprehensively reflect the actual working conditions of the battery during the charging process from different dimensions, providing richer and more specific information for model training. By determining the segment charging power and segment average temperature of the charging data segments, the abstract charging behavior is converted into quantifiable data indicators, providing a direct basis for the prediction of battery health status. In addition, using the target charging power and target average temperature of the first charging data segment as a reference benchmark, the subsequent charging segments are processed, and the change in charging power is calculated. Based on the change in charging power and the target average temperature, a multi-dimensional charging feature data set reflecting the battery health status is constructed, which eliminates the differences in initial conditions of different vehicles or different time points, ensures the comparability and consistency of the data, and thus improves the accuracy and stability of the model prediction.

[0009] Furthermore, based on the target charging power, the charging time period and the average temperature of the segments, the segment charging power is processed to obtain the charging power change of the vehicle battery in different charging time periods, including: determining the initial power change corresponding to different charging time periods based on the target charging power and the segment charging power; matching the segment average temperature with the preset temperature range to obtain a temperature matching result, and detecting the charging time period based on the preset time period to obtain a time period detection result, wherein the time detection result is used to characterize whether there are at least two charging time periods within a single preset time period; screening the initial power change based on the temperature matching result and the time period monitoring result to obtain the charging power change.

[0010] The above-described optional embodiment of the present application can achieve the following beneficial effects: By comparing the target charge capacity with the segment charge capacity, the detection system can accurately calculate the initial charge change of the battery during different charging time periods, providing an intuitive quantitative indicator for evaluating the battery's charge and discharge performance. The segment average temperature is matched with a preset temperature range to obtain a temperature matching result. The temperature matching result and the time period detection results are comprehensively analyzed to filter the initial charge change. The resulting charge change makes the charge change more consistent with actual battery usage scenarios, avoids interference from invalid or abnormal data on the prediction results, and improves the prediction accuracy of the battery status detection model.

[0011] Furthermore, based on the change in charging power and the target average temperature, charging characteristic data is constructed, including: extracting features from the change in charging power to obtain charging power characteristics generated when the vehicle battery is charged during a historical time period; and combining the charging power characteristics and the target average temperature according to a preset format to construct the charging characteristic data.

[0012] The above-mentioned optional embodiments of the present application can achieve the following beneficial effects: by extracting features from the charge level variation, the detection system can accurately capture subtle changes in battery capacity over time. By combining the target average temperature with the charge level feature, the impact of temperature changes on battery health can be quantified, providing more comprehensive input data for the prediction model. In addition, the construction of charging feature data ensures that highly relevant and high-quality data is used for model training. This not only improves the model's prediction accuracy, but also enables the model to more accurately identify vehicles at risk of low battery, reducing false alarm rates and improving the reliability of early warnings.

[0013] Furthermore, the charging characteristic data is input into a battery status detection model, and the charging characteristic data is processed using the battery status detection model to obtain a battery status of the vehicle battery, including: weighted processing of the charging characteristic data based on a preset regression coefficient to obtain a weighted characteristic value; determining a low-power probability of the vehicle battery based on the weighted characteristic value; in response to the low-power probability being greater than the preset probability, determining that the battery status is that the vehicle battery is at risk of low-power; in response to the low-power probability being less than or equal to the preset probability, determining that the battery status is that the vehicle battery does not have a low-power point risk.

[0014] The above optional embodiments of the present application can achieve the following beneficial effects: by weighting the charging feature data based on a preset regression coefficient, the detection system can identify and highlight the degree of influence of key features on the battery health status. This weighting process ensures that the model focuses more on those features that have a significant impact on the battery health status when making predictions, thereby improving the accuracy of the predictions. Based on the weighted feature values, the detection system can accurately calculate the probability of the vehicle battery being low on power. The determination of this probability is based on a large amount of historical data and machine learning model training results. It can quantify the possibility of the battery being low on power under specific usage conditions and provide a scientific basis for the assessment of the battery status. Finally, the detection system automatically classifies the battery status based on the calculated probability of low on power and the preset probability, so as to quickly determine whether the vehicle battery is at risk of low on power. This intelligent classification mechanism based on probability thresholds can promptly identify batteries in high-risk states and provide early warning information to car owners.

[0015] Furthermore, the method also includes: obtaining a sample training set of the sample battery, wherein the sample training set includes at least: sample charging data and a sample battery status of the sample battery; determining sample charging feature data of the sample battery based on the sample charging data; processing the sample charging feature data using an initial detection model to obtain an initial battery status of the vehicle battery, wherein the initial detection model includes an initial regression coefficient, and the initial regression coefficient is used to perform weighted processing on the sample charging feature data; adjusting the initial detection model based on the initial battery status and the sample battery status to obtain a battery status detection model.

[0016] The above optional embodiments of the present application can achieve the following beneficial effects: First, charging feature data are extracted from the charging data of the sample battery. These feature data reflect the performance degradation pattern of the battery under actual use environment, providing high-quality input for subsequent model training. Subsequently, the extracted charging feature data are processed using an initial detection model including an initial regression coefficient. The model can more sensitively capture subtle changes in the feature data and preliminarily predict the initial state of the vehicle battery, laying the foundation for subsequent training of the model. Finally, based on the comparison between the battery state predicted by the initial model and the actual sample battery state, the initial detection model can be adjusted to obtain a battery state detection model, so that the battery state detection model can more accurately simulate the battery aging process, reduce prediction errors, and improve prediction accuracy.

[0017] Furthermore, the initial detection model is adjusted based on the initial battery state and the sample battery state to obtain a battery state detection model, including: constructing a regression coefficient loss function based on the initial battery state and the sample battery state; adjusting the initial regression coefficient based on the regression coefficient loss function to obtain a preset regression coefficient; and constructing a battery state detection model based on the preset regression coefficient and the initial detection model.

[0018] The above optional embodiment of the present application can achieve the following beneficial effects: by comparing the initial battery state with the sample battery state, a regression coefficient loss function is constructed. This function measures the gap between the model prediction result and the actual battery state, and is an important tool for adjusting model parameters to optimize the prediction effect. Based on the above loss function, the detection system can automatically adjust the initial regression coefficient until the expected state is reached, obtain the preset regression coefficient, and construct a battery state detection model based on the regression coefficient and the initial detection model. This process continuously adjusts the regression coefficient through iterative calculation, so that the constructed battery state detection model can more accurately capture the inherent laws of battery state changes during training, so that the battery health status and power outage risk can be more accurately assessed in the prediction stage, thereby improving the accuracy of the prediction.

[0019] Furthermore, historical charging data of the vehicle battery on the target vehicle in the historical time period is obtained, including: obtaining the power change data and historical power of the vehicle battery in the historical time period, as well as the operating status of the vehicle battery; based on the operating status and historical power, initial charging data is filtered from the power change data; data cleaning is performed on the initial charging data to obtain cleaned charging data; missing values are filled in the cleaned charging data to obtain historical charging data.

[0020] The above optional embodiments of the present application can achieve the following beneficial effects: First, the detection system collects the power change data and historical power values of the vehicle battery within the historical time period. This step ensures that we can grasp the full picture of the power fluctuations during the battery life cycle, and provide rich and detailed data support for subsequent health predictions. Subsequently, based on the vehicle's operating status and historical power information, the detection system can accurately filter out data fragments related to power replenishment from the power change data, namely the initial power replenishment data. This step eliminates power changes during non-power replenishment periods, reduces the interference of irrelevant data, and improves the pertinence and efficiency of subsequent analysis. After obtaining the initial power replenishment data, the detection system cleans the initial power replenishment data, and further fills the missing values in the cleaned power replenishment data to obtain historical power replenishment data, thereby improving the integrity and accuracy of the historical power replenishment data, and further improving the training effect of the initial detection model.

[0021] Furthermore, the method also includes: analyzing a user profile of a user who owns the target vehicle to determine the user's information acquisition characteristics, wherein the information acquisition characteristics are used to reflect the user's information acquisition habits; in response to the battery status indicating that the vehicle battery is at risk of low power, generating an information push method based on the information acquisition characteristics; and pushing risk information to the user based on the information push method, wherein the risk information is used to prompt the user that the vehicle battery is at risk of low power.

[0022] The above optional embodiments of the present application can achieve the following beneficial effects: through in-depth analysis of the user portraits of the target vehicles, the system can accurately grasp the information acquisition habits of each user, which is the basis for achieving accurate information push. Based on the information acquisition characteristics obtained from the analysis, the detection system intelligently generates a warning information push method that is more suitable for the user, ensuring that the warning information can be delivered to the user quickly and accurately, reducing the risk of information omission or delay. When it is determined that the vehicle battery is at risk of low power, the system immediately sends risk warning information to the user according to a customized push method. This instant communication mechanism can promptly awaken users to pay attention to battery health, prompting them to take necessary inspection or maintenance actions, and avoid driving inconvenience or safety hazards caused by battery problems.

[0023] According to one embodiment of the present application, a battery status detection device is provided, comprising: a data acquisition module for acquiring historical charging data of a vehicle battery on a target vehicle during a historical time period, wherein the historical charging data is used to represent data generated when a power supply battery supplies power to a vehicle battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery; a fragment interception module for performing data analysis on the historical charging data and intercepting multiple charging data fragments from the historical charging data, wherein the multiple charging data fragments are used to represent data fragments of continuous charging of the vehicle battery within a preset charging current range; a feature extraction module for performing feature extraction on the multiple charging data fragments to obtain charging feature data of the vehicle battery, wherein the charging feature data is used to reflect changes in the charge level and temperature of the vehicle battery when the target vehicle is charged during the historical time period; and a status detection module for inputting the charging feature data into a battery status detection model, processing the charging feature data using the battery status detection model, and obtaining a battery status of the vehicle battery, wherein the battery status is used to represent whether the vehicle battery is at risk of power failure.

[0024] Furthermore, the feature extraction module is also used to: perform feature extraction on multiple charging data segments to obtain charging feature data of the vehicle battery, including: performing feature extraction on the charging data segments to obtain the charging time period and charging parameters corresponding to the charging data segments; based on the charging parameters, determining the segment charging power and segment average temperature corresponding to the charging data segments, wherein the segment charging power is used to characterize the charging power of the vehicle battery in the corresponding charging time period, and the segment average temperature is used to characterize the average temperature of the vehicle battery in the corresponding charging time period; processing the segment charging power based on the target charging power, the charging time period and the segment average temperature to obtain the charging power change of the vehicle battery in different charging time periods, wherein the target charging power is used to characterize the segment charging power corresponding to the first charging data segment among the multiple charging data segments; constructing the charging feature data based on the charging power change and the target average temperature, wherein the target average temperature is used to characterize the segment average temperature corresponding to the first charging data segment among the multiple charging data segments.

[0025] Furthermore, the feature extraction module is also used to: determine the initial power change corresponding to different charging time periods based on the target charging power and the segment charging power; match the average temperature of the segment with the preset temperature range to obtain a temperature matching result, and detect the charging time period based on the preset time period to obtain a time period detection result, wherein the time detection result is used to characterize whether there are at least two charging time periods within a single preset time period; screen the initial power change based on the temperature matching result and the time period monitoring result to obtain the charging power change.

[0026] Furthermore, the feature extraction module is also used to: extract features from the charging power change to obtain the charging power features generated when the vehicle battery is recharged during a historical time period; combine the charging power features and the target average temperature according to a preset format to construct recharge feature data.

[0027] Furthermore, the status detection module is also used to: perform weighted processing on the charging characteristic data based on a preset regression coefficient to obtain a weighted characteristic value; determine the probability of the vehicle battery being out of power based on the weighted characteristic value; in response to the probability of the power being out of power being greater than the preset probability, determine that the battery status is that the vehicle battery is at risk of being out of power; in response to the probability of the power being out of power being less than or equal to the preset probability, determine that the battery status is that the vehicle battery does not have a risk of being out of power.

[0028] Furthermore, the device also includes: a first acquisition module, used to obtain a sample training set of a sample battery, wherein the sample training set includes at least: sample charging data and a sample battery status of the sample battery; a first determination module, used to determine the sample charging feature data of the sample battery based on the sample charging data; a first processing module, used to process the sample charging feature data using an initial detection model to obtain an initial battery status of the vehicle battery, wherein the initial detection model includes an initial regression coefficient, and the initial regression coefficient is used to perform weighted processing on the sample charging feature data; a first adjustment module, used to adjust the initial detection model based on the initial battery status and the sample battery status to obtain a battery status detection model.

[0029] Furthermore, the first adjustment module is also used to: construct a regression coefficient loss function based on the initial battery state and the sample battery state; adjust the initial regression coefficient based on the regression coefficient loss function to obtain a preset regression coefficient; and construct a battery state detection model based on the preset regression coefficient and the initial detection model.

[0030] Furthermore, the data acquisition module is also used to: obtain the power change data and historical power of the vehicle battery in a historical time period, as well as the operating status of the vehicle battery; based on the operating status and historical power, filter the initial power replenishment data from the power change data; perform data cleaning on the initial power replenishment data to obtain cleaned power replenishment data; fill in missing values in the cleaned power replenishment data to obtain historical power replenishment data.

[0031] Furthermore, the device also includes: a second determination module, used to analyze the user portrait of the user who owns the target vehicle and determine the user's information acquisition characteristics, wherein the information acquisition characteristics are used to reflect the user's information acquisition habits; a second acquisition module, used to generate an information push method based on the information acquisition characteristics in response to the battery status indicating that the vehicle battery is at risk of low power; a first push module, used to push risk information to the user based on the information push method, wherein the risk information is used to prompt the user that the vehicle battery is at risk of low power. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of a battery status detection method provided by an embodiment of the present application;

[0033] Figure 2 This is a flow chart of vehicle battery health prediction provided by an embodiment of the present application;

[0034] Figure 3 is a structural diagram of a battery status detection device provided in an embodiment of the present application;

[0035] Figure 4 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] In order to make the technical problems, technical solutions and beneficial effects solved by this application more clearly understood, this application is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0037] As new energy vehicles age, the health of batteries, a crucial component of new energy vehicle systems, directly impacts their reliability and safety. Once a battery enters its decline phase, its storage capacity and output power decrease significantly, potentially preventing the vehicle from starting or even causing a safety accident. Currently, the detection and prediction of the health of new energy vehicle batteries relies primarily on the following methods: Direct measurement: This involves directly measuring the remaining capacity by regularly performing deep discharge tests on the battery. However, this method is destructive and not suitable for routine testing. Current integration: This method uses the integration of battery charge and discharge current data during vehicle operation to estimate the remaining capacity. However, this method is affected by the accuracy of the current sensor and changes in vehicle load, resulting in large errors in the estimated remaining capacity.

[0038] Some nouns or terms that appear in the description of the embodiments of this application are subject to the following interpretations:

[0039] Direct Current to Direct Current Converter (DC-DC): A DC-DC converter is an electronic device used to convert the voltage of one DC power source to a different DC voltage. These converters are widely used in various power electronics systems, including but not limited to computer power supplies, communications equipment, automotive electronics, portable electronic devices, and aerospace systems.

[0040] The Electronic Battery System Code (EBSCode) is a set of numbers and letters with specific information that identify the main attributes and uniqueness of the power battery. The coding targets are automotive power battery packs, battery modules, single cells, and power battery packs, battery modules, and single cells for cascade utilization. The codes of power battery packs, battery modules, and single cells, as well as power battery packs, battery modules, and single cells for cascade utilization, should correspond to each other.

[0041] A battery status detection method provided in an embodiment of the present application includes: obtaining historical charging data of a vehicle battery in a historical time period, wherein the historical charging data is used to represent data generated when a power supply battery supplies power to a vehicle battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery; performing data analysis on the historical charging data, and intercepting multiple charging data segments from the historical charging data, wherein the multiple charging data segments are used to represent data segments of continuous charging of the vehicle battery within a preset charging current range; performing feature extraction on the multiple charging data segments to obtain charging characteristic data of the vehicle battery, wherein the charging characteristic data is used to reflect the changes in the power level and temperature of the vehicle battery when the target vehicle is charged within the historical time period; and processing the charging characteristic data using a battery status detection model to obtain a battery status of the vehicle battery, wherein the battery status is used to represent whether the vehicle battery is at risk of power outage.

[0042] The above-mentioned battery status detection method provided in the embodiment of the present application achieves the following technical effects: by obtaining the historical charging data of the vehicle battery in the historical time period, the pertinence and effectiveness of the data are ensured; by conducting in-depth analysis of the historical charging data, continuous charging data fragments within the preset charging current range are intercepted, and high-quality charging data are further screened out, and interference data of abnormal current or discontinuous charging are excluded; then, key charging feature data is extracted from the charging data fragments, and the original data is converted into feature data with more analytical value, making the data easier to be understood and processed by the model; finally, the charging feature data is processed by the battery status detection model, and the intrinsic connection between the charging feature data and the battery health status is learned by training the model, thereby obtaining the battery status of the vehicle battery, thereby achieving the purpose of accurately identifying whether the vehicle battery has the risk of low power, and achieving the technical effect of improving the prediction accuracy of the vehicle battery health status, thereby solving the technical problem of low accuracy in predicting the vehicle battery health status in the prior art.

[0043] Example 1

[0044] This application embodiment provides a battery status detection method, please refer to Figure 1 , including the following steps:

[0045] S102: Acquire historical charging data of the vehicle battery in a historical time period, wherein the historical charging data is used to represent data generated when the power supply battery supplies power to the vehicle battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery.

[0046] The historical time period mentioned above can refer to a longer period of time in the past during vehicle use. This period is typically selected to collect sufficient vehicle operating data for analysis and prediction. The length of the historical time period can be set based on specific needs, such as weeks, months, or even years, to ensure that the collected data covers the vehicle battery's performance under various operating conditions.

[0047] The above-mentioned historical power replenishment data may be a data record generated when the power supply battery supplements the vehicle battery during the vehicle's usage history. The above-mentioned power replenishment may refer to the process in which the system automatically or manually triggers the high-voltage battery to charge the low-voltage battery through a DC-DC converter when the voltage or power of the vehicle battery is lower than a certain threshold.

[0048] In an optional embodiment, considering that the driving conditions of the vehicle during driving are relatively complex, if the charging data is directly obtained during the actual use of the vehicle, there may be large errors. Therefore, the staff can pre-equip the vehicle with a vehicle network terminal that can record the charging process data. The terminal can record the charging data of the vehicle battery in real time and periodically send the data to the battery status detection system (hereinafter referred to as the detection system), so that the detection system can periodically collect the charging data of the vehicle battery. After completing the data collection, the detection system can filter out data segments that meet the charging conditions from a large amount of data as the above-mentioned historical charging data.

[0049] For example, the raw data in the power-up segment can be expressed as follows:

[0050] Data[X,Y];

[0051] X∈[0,n-1];

[0052] Y∈[Y0,Y m-1 ].

[0053] In the formula, X represents n data, [0,n-1] represents the above n data, Y represents that one acquisition cycle consists of m signals, [Y0,Y m-1] represents the m signals, where the signals may include at least one or more of the following: vehicle code, battery code, collection time, vehicle status, door status, battery voltage, battery temperature, battery current, etc., but are not limited thereto. The filtering conditions for the historical charging data may be: the new energy vehicle is in the off state, four doors and two lids are closed, the high-voltage battery is powered, the DC-DC output voltage is normal, the battery current is positive, and the vehicle speed is zero.

[0054] In another optional embodiment, the detection system can use sensors to obtain the charging data of the vehicle battery for all time periods in real time when the vehicle is stationary, and store the data and the timestamp corresponding to the data in a pre-built database, so that the detection system can directly read the charging data from the database when it needs to obtain historical charging data. Specifically, the detection system can pre-divide the historical time period that needs to be read, and then read the charging data corresponding to the specified historical time period from the database based on the timestamp, and use the charging data as the historical charging data.

[0055] S104: Analyze the historical charging data and extract multiple charging data segments from the historical charging data, wherein the multiple charging data segments are used to represent data segments of continuous charging of the vehicle battery within a preset charging current range.

[0056] The above-mentioned charging data segment may refer to a data record within a continuous period of time intercepted from the historical charging data. The above-mentioned preset charging current range may be a charging current range pre-set according to the battery characteristics and charging requirements.

[0057] In an optional embodiment, considering that selecting data segments within a preset charging current range can ensure that data is collected under similar charging conditions, thereby reducing noise introduced by different charging strategies or abnormal operating conditions and improving data consistency and accuracy, the detection system can pre-set a charging current range based on battery characteristics and charging strategies. For example, the detection system can determine a charging current interval of 5A to 20A, which helps to highlight those charging processes that have a significant impact on battery performance, while ignoring those data where the current is too low to produce a significant effect or the current is too high to cause anomalies. Subsequently, the detection system can define the criteria for charging data segments. For example, a segment can start when the charging current first enters the preset range and end when the current drops below the lower limit of the preset range. Then, based on the above criteria, the detection system can use a sliding window or fixed time interval method to segment the historical charging data into multiple independent data segments, ensuring that each segment represents a continuous charging process within the preset charging current range.

[0058] For example, in order to cover batteries in different health states, the detection system can intercept data segments with the same current range and continuous current according to the vehicle code as the above data segments. Specifically, the above data segments can be expressed as follows:

[0059] D=[I1,I2,…,I n ].

[0060] Where D represents the data segment, I1 represents the first frame of the selected interval data, and I n is the termination current within the selected interval.

[0061] It should be noted that the specific values of the above-mentioned supplementary current range are only for illustrative purposes. Staff can set them according to actual needs and are not limited here.

[0062] S106: Extract features from the multiple charging data segments to obtain charging feature data of the vehicle battery, wherein the charging feature data is used to reflect the changes in power level and temperature of the vehicle battery when charging the target vehicle within a historical time period.

[0063] The above-mentioned charging characteristic data may be data extracted from multiple charging data segments, and is used to describe the changes in key performance parameters such as power and temperature of the vehicle battery during the charging process within a historical time period. The above-mentioned power change situation may be the change in the power of the target vehicle battery from the beginning to the end of the charging process. For example, the above-mentioned power change situation may include at least one or more of the following: power before charging, power after charging, power difference, power change trend, etc., but is not limited to these. The above-mentioned temperature change situation may be the change in the temperature of the target vehicle battery during the charging process. For example, the above-mentioned temperature change situation may include at least one or more of the following: temperature before charging, average temperature during charging, temperature change range during charging, temperature change rate, etc., but is not limited to these.

[0064] In an optional embodiment, considering that the original multiple charging data fragments may contain a large amount of redundant information, such as timestamps and vehicle status, in order to reduce the data volume, refine the information, and reduce the complexity of subsequent processing and model training, the detection system can process the above multiple charging data fragments through feature extraction to obtain charging feature data, so that the above charging feature data includes power change and temperature change indicators, and further enables the above charging feature data to directly reflect the performance and status of the vehicle battery during the charging process.

[0065] S108: Process the charging characteristic data using a battery status detection model to obtain a battery status of the vehicle battery, wherein the battery status is used to indicate whether the vehicle battery is at risk of low power.

[0066] The above-mentioned battery status detection model can be a prediction model based on machine learning or deep learning, which aims to predict and evaluate the health status of the battery and the potential risk of power shortage by analyzing the charging feature data. The above-mentioned battery status detection model can be trained by a large amount of historical battery operation data, and can capture and learn the characteristic performance of the battery in different states. For example, the above-mentioned characteristics may include at least one or more of the following: charging efficiency, temperature effect, voltage change and other parameters related to battery health, etc., but not limited to these. The above-mentioned battery status detection model can adopt a variety of algorithms to adapt to different types of data and prediction needs. For example, the above-mentioned algorithms may include at least one or more of the following: logistic regression, support vector machine, random forest, neural network, etc., but not limited to these.

[0067] The battery status may be a status indicating the health of the vehicle battery and whether there is a risk of low battery. For example, the battery status may include at least one or more of the following: healthy status, low battery status, etc., but is not limited thereto.

[0068] In an optional embodiment, considering that the battery status is an important indicator for evaluating battery performance, in order to enable the detection system to understand whether the battery is in a normal state of charge or is in a state of low charge, the detection system can use the above-mentioned battery status detection model to analyze the charging feature data, thereby obtaining the battery status of the vehicle battery, so that the detection system can identify the risk of low charge of the vehicle battery as early as possible. Specifically, the detection system can pre-select or train a battery status detection model, which can be a machine learning model (such as logistic regression, support vector machine, random forest, neural network, etc.). Subsequently, the detection system can input the above-mentioned charging feature data into the model. The model will process and analyze the input feature data according to the rules or algorithms it has learned (such as logistic regression, support vector machine, random forest, neural network, etc.), and output the battery status of the vehicle battery. Based on the above-mentioned battery status, the detection system can evaluate whether the vehicle battery is at risk of low charge.

[0069] Based on the above steps S102 to S108, by obtaining the historical charging data of the vehicle battery within the historical time period, the pertinence and effectiveness of the data are ensured. By conducting in-depth analysis of the historical charging data, continuous charging data segments within the preset charging current range are intercepted, and high-quality charging data are further screened out, eliminating interference data of abnormal current or discontinuous charging. Subsequently, key charging feature data are extracted from the charging data segments, and the original data is converted into feature data with more analytical value, making the data easier to be understood and processed by the model. Finally, the charging feature data is processed using the battery status detection model, and the intrinsic relationship between the charging feature data and the battery health status is learned by training the model, thereby obtaining the battery status of the vehicle battery, thereby achieving the purpose of accurately identifying whether the vehicle battery has a risk of low power, and realizing the technical effect of improving the prediction accuracy of the vehicle battery health status, thereby solving the technical problem of low accuracy in predicting the vehicle battery health status in the prior art.

[0070] Furthermore, feature extraction is performed on multiple power-recharging data segments to obtain power-recharging characteristic data of the vehicle battery, including: feature extraction is performed on the power-recharging data segments to obtain the power-recharging time period and power-recharging parameters corresponding to the power-recharging data segments; based on the power-recharging parameters, the segment charging power and segment average temperature corresponding to the power-recharging data segments are determined, wherein the segment charging power is used to characterize the charging power of the vehicle battery in the corresponding power-recharging time period, and the segment average temperature is used to characterize the average temperature of the vehicle battery in the corresponding power-recharging time period; the segment charging power is processed based on the target charging power, the power-recharging time period and the segment average temperature to obtain the charging power change of the vehicle battery in different power-recharging time periods, wherein the target charging power is used to characterize the segment charging power corresponding to the first power-recharging data segment among the multiple power-recharging data segments; based on the charging power change and the target average temperature, the power-recharging characteristic data is constructed, wherein the target average temperature is used to characterize the segment average temperature corresponding to the first power-recharging data segment among the multiple power-recharging data segments.

[0071] The aforementioned charging time period can be used to determine how many times per week charging data is recorded. The aforementioned charging parameters can be signals generated during the charging process. For example, these charging parameters can include at least one or more of the following: charging current, battery temperature, and battery voltage, but are not limited to these. These charging parameters are extracted from the charging data segments for subsequent analysis. The aforementioned segment charge capacity can be the total charge received by the battery during the current charging time period, i.e., the increase in battery capacity from the start to the end of charging. The aforementioned segment average temperature can be the average temperature of the vehicle battery calculated during each charging time period. This segment average temperature can be used to assess the battery's temperature during the charging process. The aforementioned target charge capacity can be the segment charge capacity corresponding to the first charging data segment. By comparing the charge capacities of other segments with the target charge capacity, it is possible to assess whether the battery's charging capacity has degraded over time. The aforementioned target average temperature can be the segment average temperature corresponding to the first charging data segment. This temperature is used as a temperature benchmark in subsequent analysis to help the detection system understand changes in battery performance under different temperature conditions.

[0072] In an optional embodiment, considering that analyzing differences in battery performance across different time periods (e.g., morning and evening, and seasonal variations) helps the detection system identify patterns and anomalies in battery behavior, the detection system can also quantify battery performance during each recharging process by analyzing the aforementioned recharging parameters. Therefore, the detection system can first perform feature extraction on the aforementioned recharging data segments to obtain the corresponding recharging time periods and recharging parameters. These recharging parameters may include data such as recharging time, vehicle battery current, and vehicle battery temperature. Based on these recharging parameters, the detection system can further calculate the vehicle battery charge level and average battery temperature during the corresponding recharging time period, i.e., the segment charge level and segment average temperature. To eliminate the impact of initial differences between different vehicle batteries, the detection system can further process the segment charge levels based on the charge level of the first recharging data segment, i.e., the target charge level, as well as the recharging time period and segment average temperature, to calculate the change in charge level of the vehicle battery during different time periods. Finally, the detection system can construct the recharging characteristic data based on the change in charge level and the average temperature of the first recharging data segment, i.e., the target average temperature.

[0073] Specifically, the detection system may first calculate data such as the charge capacity and average temperature based on the selected data segments. Specifically, the charge capacity of the battery within the same charging current range may be calculated as shown in the following formula:

[0074]

[0075] Where Q represents the charging capacity, I irepresents the battery current, t represents the charging time, and n represents the amount of interval data. The average charging temperature of the battery can be expressed as follows:

[0076]

[0077] Where, T avg Indicates the average temperature of the battery, T i represents the real-time collected temperature value of the battery. The other symbols in the formula have the same meanings as in the previous formula and are not repeated here. After completing the above calculations of charge capacity and average temperature, the detection system can use the vehicle code as an index to establish a data set of battery charge capacity decay over time. This data set can be shown as follows:

[0078] {vin,EBSCode,[Time1,Q1,T avg1 ],[Time2,Q2,T avg2 ],…[Time i ,Q i ,T avgi ]…[Time n ,Q n ,T avgn ]};

[0079] In the formula, vin represents vehicle code, EBSCode represents battery code, Time i Indicates the timestamp or time point after the i-th recharge, Q i It represents the charging capacity of the vehicle battery after the i-th charging, T avgi represents the average temperature of the vehicle battery after the i-th recharge. The other symbols have the same meanings as in the previous formula and are not repeated here. After the data set is constructed, the detection system can use big data statistical characteristics (such as mean, standard deviation, maximum, minimum, etc.) as battery health factors, extract characteristic values, and obtain the above-mentioned recharge characteristic data. The above-mentioned recharge characteristic data can be expressed as follows:

[0080] {vin,ΔQ max1 ,ΔQ min1 ,ΔQ avg1 ,ΔQ std1 ,ΔQ1,T avg1};

[0081] Where ΔQ max1 It represents the maximum value of the difference in the vehicle battery charge during the evaluation cycle, that is, the maximum drop in the vehicle battery's charging capacity during the cycle compared to the first recharge, ΔQ min1 Indicates the minimum difference in the vehicle battery charge during the evaluation cycle, that is, the minimum drop in the vehicle battery's charging capacity during the cycle relative to the first recharge, ΔQavg1 Indicates the average value of the difference in vehicle battery charge during the evaluation period, ΔQ std1 It represents the standard deviation of the difference in vehicle battery charge during the evaluation period. The other symbols have the same meanings as in the previous formula and are not repeated here.

[0082] Furthermore, based on the target charging power, the charging time period and the average temperature of the segments, the segment charging power is processed to obtain the charging power change of the vehicle battery in different charging time periods, including: determining the initial power change corresponding to different charging time periods based on the target charging power and the segment charging power; matching the segment average temperature with the preset temperature range to obtain a temperature matching result, and detecting the charging time period based on the preset time period to obtain a time period detection result, wherein the time detection result is used to characterize whether there are at least two charging time periods within a single preset time period; screening the initial power change based on the temperature matching result and the time period detection result to obtain the charging power change.

[0083] The aforementioned initial charge change may refer to the charge change obtained by comparing the vehicle battery's charge level with the target charge level (i.e., the charge level corresponding to the first charge data segment) during different charge time periods. The aforementioned temperature matching result may be a comparison of the average temperature of the segments in the charge data segment with a preset temperature range to determine whether the charge event was performed under suitable temperature conditions. The suitable temperature range is generally set to ensure battery charging efficiency and extend battery life. The aforementioned time period detection result may be a result used to detect whether multiple charge time periods exist within a single preset time period (e.g., one week).

[0084] In an optional embodiment, considering that by comparing the charge level after each recharge with the charge level after the first recharge, differences in initial battery status can be eliminated and the accuracy of detection results improved, the detection system can calculate the difference between the target charge level and the segmented charge level to determine the initial charge change corresponding to different time periods. In addition to eliminating differences in initial battery status, battery performance is highly dependent on operating temperature. Extremely low or high temperatures can affect battery charging efficiency and performance. Therefore, by matching the data with a preset optimal temperature range, the detection system can filter out recharge data segments operating under normal temperature conditions, thereby improving the accuracy of detection results. Furthermore, by detecting whether there are multiple recharge events within a preset time period, the detection system can identify vehicles whose batteries frequently require recharges. Since frequent recharges may be associated with poor battery health, data from vehicles that frequently recharge within the preset data period better reflects the actual battery condition than data from vehicles that only recharge occasionally, making the final detection results more realistic and more reliable. Finally, to ensure that the data input into the vehicle status detection model in subsequent steps is at an appropriate temperature and represents a typical battery usage pattern, the detection system can combine the temperature matching results and time period detection results to filter out the initial power change data that meets the conditions, thereby obtaining the above-mentioned charging power change.

[0085] For example, considering that the starting point Q1 of each vehicle's high-voltage battery recharging the low-voltage battery is random, in order to standardize the data and eliminate the impact of differences in the vehicle's initial charge state on subsequent analysis, thereby better evaluating and predicting the battery's health, the detection system can standardize the charge data as shown in the following formula:

[0086] ΔQ i =Q1-Q i ;

[0087] Where ΔQ i represents the normalized charge level of the vehicle battery after the i-th recharge. Based on the above formula, by calculating the difference between each charge level and the initial charge level, the detection system can normalize the charge level data relative to the initial state of the vehicle battery, eliminating differences in the initial charge levels of different vehicles and allowing all vehicle data to be compared and analyzed on the same basis. After completing the above charge level data standardization, further considering that the battery temperature affects the charging current, in order to improve the accuracy of the charging current data, the detection system can also use T avgThe value of can filter out data with too high or too low temperature. For example, the detection system can eliminate data with battery temperature greater than 70℃ and below 0℃. In addition, it is also considered that the charging process of the vehicle battery will be interfered by various factors, such as power grid fluctuations, user behavior, temperature changes, etc. These factors may cause certain fluctuations and anomalies in the charging data. By selecting the maximum charging amount per week, the data noise generated by these short-term interferences can be reduced, making the data set more stable and representative. Therefore, if the above-mentioned vehicle has multiple charging data records within a week, the detection system can select the maximum charging amount in the weekly charging data to form a new data set. The new data can be shown as follows:

[0088] {vin,ΔQ1,ΔQ2,…,ΔQ i ,…,ΔQ m};

[0089] Where ΔQ m It represents the difference between the charge amount of the vehicle during the mth recharge and the charge amount of the vehicle during the first recharge. The meanings of other symbols in the formula are the same as those in the previous formula and are not repeated here.

[0090] Furthermore, based on the change in charging power and the target average temperature, charging characteristic data is constructed, including: extracting features from the change in charging power to obtain charging power characteristics generated when the vehicle battery is charged during a historical time period; and combining the charging power characteristics and the target average temperature according to a preset format to construct the charging characteristic data.

[0091] The aforementioned charging power characteristics may be extracted from historical charging data, and are intended to reflect the trends and patterns of changes in the vehicle battery power during charging over different time periods. The aforementioned preset format may be a format that organizes the charging power characteristics and target average temperature according to certain rules and structures to facilitate subsequent data processing and machine learning model training.

[0092] In an optional embodiment, considering that the charge capacity variation is important data for vehicle battery testing, but this data may contain noise or exhibit nonlinear relationships, making it difficult to be directly utilized by the battery status detection model, the detection system can convert the charge capacity variation into more meaningful features through feature extraction, such as statistical data such as the mean, standard deviation, maximum, and minimum values of the charge capacity variation. Subsequently, the detection system can construct charge capacity characteristics generated by the vehicle battery during recharging over a historical time period based on these statistical data. These characteristics can more comprehensively describe the performance trend of the battery during multiple recharging processes. After obtaining the charge capacity characteristics, further considering that battery performance is significantly affected by temperature, even under the same charging conditions, different temperatures can cause significant changes in battery charging efficiency and capacity. Therefore, the detection system can combine the charge capacity characteristics with the target average temperature to construct the recharging characteristic data, ensuring that the battery status detection model can account for the temperature dependence of battery performance. This combined characteristic data can more comprehensively reflect the operating environment and conditions of the battery, thereby improving the prediction accuracy and reliability of the battery status detection model.

[0093] For example, the detection system can establish a sample input of an increasing sequence of charging capacity for each vehicle, which can be shown as follows:

[0094] x(t i )={vin,EBSCode,ΔQ max ,ΔQ min ,ΔQ avg ,ΔQ std ,ΔQ}.

[0095] Where ΔQ max Indicates the maximum value of the change in charge, ΔQ min Indicates the minimum value of the change in charge capacity, ΔQ avg Indicates the average value of the change in charge capacity, ΔQ std represents the standard deviation of the change in charge capacity, ΔQ represents the change in charge capacity, and the meanings of other symbols are the same as those in the previous formula and will not be repeated here.

[0096] Furthermore, the charging characteristic data is input into a battery status detection model, and the charging characteristic data is processed using the battery status detection model to obtain a battery status of the vehicle battery, including: weighted processing of the charging characteristic data based on a preset regression coefficient to obtain a weighted characteristic value; determining a low-power probability of the vehicle battery based on the weighted characteristic value; in response to the low-power probability being greater than the preset probability, determining that the battery status is that the vehicle battery is at risk of low-power; in response to the low-power probability being less than or equal to the preset probability, determining that the battery status is that the vehicle battery does not have a low-power point risk.

[0097] In an optional embodiment, considering that in the machine learning model, the regression coefficient represents the degree of influence of each feature on the prediction result, through weighted processing, the battery status detection model can adjust the importance of the feature value according to these coefficients, which means that when predicting the risk of power outage, the model will pay more attention to those features that have a greater impact on the battery health status, thereby improving the accuracy of the model's prediction. Therefore, the detection system can perform weighted processing on the charging feature data based on the above-mentioned preset regression coefficients to obtain weighted feature values. The above-mentioned weighted feature values can be directly used in the battery status detection model to determine the probability of power outage of the vehicle battery. For example, the above-mentioned battery status detection model can be a logistic regression model, and the above-mentioned weighted feature values can be used to calculate the probability of power outage events. The value of the above-mentioned probability can be a continuous value between 0 and 1, indicating the possibility of power outage of the vehicle battery under specific conditions. After calculating the above-mentioned probability of low battery, the detection system can determine whether the current vehicle battery has a low battery risk based on the size relationship between the probability value and the preset probability threshold. For example, if the above-mentioned probability of low battery is greater than the preset threshold, the detection system can consider that the current vehicle battery has a low battery risk. Correspondingly, if the above-mentioned probability of low battery is less than the preset threshold, the detection system can consider that the current vehicle battery does not have a low battery risk.

[0098] For example, the detection system can pre-build a logistic regression model as the above-mentioned battery status detection model. The logic of the model can be shown as follows:

[0099] ln(P / 1-P)=β0+β1X1+β2X2+…+β n X n ;

[0100] Where P is the probability of a positive sample, that is, the probability of power failure, X n represents the nth feature, β0, β1, ... β n Represents the intercept term and the regression coefficient corresponding to each feature, β0, β1, ... β n It can be solved by maximum likelihood estimation. After completing the construction of the above logistic regression model, the detection system can input the above feature data into the model. Since ln(P / 1-P)>0, then P / 1-P>1, based on this, P>0.5, therefore, the vehicle data to be predicted is substituted into the above logistic regression model to determine β0+β1X1+β2X2+…+β n X n Is the value of greater than 0? If so, it means that the probability P of the current vehicle being out of power is greater than 0.5. If not, it means that the probability P of the current vehicle being out of power is less than 0.5. For vehicles with a probability of power outage greater than 0.5, the above logistic regression model can be considered that the current vehicle is out of power or has a risk of power outage.

[0101] It should be noted that the above-mentioned specific values such as the probability of power failure are only shown for example. The staff can set them according to actual needs and are not limited here.

[0102] Furthermore, the method also includes: obtaining a sample training set of the sample battery, wherein the sample training set includes at least: sample charging data and a sample battery status of the sample battery; determining sample charging feature data of the sample battery based on the sample charging data; processing the sample charging feature data using an initial detection model to obtain an initial battery status of the vehicle battery, wherein the initial detection model includes an initial regression coefficient, and the initial regression coefficient is used to perform weighted processing on the sample charging feature data; adjusting the initial detection model based on the initial battery status and the sample battery status to obtain a battery status detection model.

[0103] The above-mentioned sample charging data can be extracted from the historical charging records of vehicle batteries in known healthy and low-power states, and used as a data set for training and optimizing the battery status detection model. The above-mentioned sample battery status can be the actual health status of the sample battery at a certain moment in history. The above-mentioned sample battery status can be determined by various means such as expert experience, market feedback, laboratory testing, etc., and used as a label for training the battery status detection model. The above-mentioned initial detection model can be a model designed based on preliminary set parameters and algorithms, which is used to make a preliminary prediction of the above-mentioned sample charging feature data. The above-mentioned initial battery status can be the vehicle battery status predicted by the above-mentioned initial detection model based on the sample charging feature data.

[0104] In an optional embodiment, considering that the sample training set is the training basis of the initial detection model, it contains the battery charging data and actual battery status (whether it is low on power or there is a risk of low on power) in historical data. Therefore, the detection system can first collect the charging data and related status information of the sample batteries from the Internet of Vehicles big data, and label the status of each sample battery (such as whether it is low on power) based on historical records or expert opinions, thereby constructing a sample training set of the above-mentioned sample batteries. The sample training set at least includes the sample charging data and sample battery status of the above-mentioned sample batteries, so that the initial detection model can learn the association between the battery status and charging characteristics from the sample data set. After obtaining the above sample data set, further considering that the original sample data set often contains noise and has various forms, in order to improve the training effect of the initial detection model, the detection system can preprocess the collected sample data set, such as cleaning, filling missing values, and checking for outliers. Then, the detection system can extract features from the preprocessed data, such as the mean, standard deviation, maximum and minimum values of the charge capacity, and the average temperature during charging. Subsequently, the detection system can determine which features are helpful in predicting the battery state through statistical tests or feature importance analysis, thereby determining the sample charging feature data. After determining the above sample charging feature data, the detection system can select a logistic regression model as the initial detection model and initialize the parameters of the model. Subsequently, the detection system can use the above initial detection model to process the sample charging feature data, perform weighted summation using the current regression coefficient, and then transform it through the activation function to obtain the initial battery state. Finally, the detection system can calculate the model loss or error by comparing the initial battery state predicted by the initial detection model with the actual sample battery state, and adjust the initial detection model based on the calculated error or loss to obtain the battery state detection model.

[0105] For example, the detection system can construct the above-mentioned sample training set in the following way: based on the vehicle's most recent charging time period, select vehicle data for the past week, and calculate whether the battery voltage of the vehicle is lower than a preset voltage threshold in a continuous time period in the recent period. Then, based on the charging capacity statistics of the same battery model and laboratory aging constant voltage charging data, market feedback on power shortage data, etc., finally, according to the expert data labeling method, the risk identification LBindex of the vehicle in the sample with power shortage is marked, LBindex = 1 indicates power shortage and power shortage risk, LBindex = 0 indicates no power shortage and power shortage risk.

[0106] Furthermore, the initial detection model is adjusted based on the initial battery state and the sample battery state to obtain a battery state detection model, including: constructing a regression coefficient loss function based on the initial battery state and the sample battery state; adjusting the initial regression coefficient based on the regression coefficient loss function to obtain a preset regression coefficient; and constructing a battery state detection model based on the preset regression coefficient and the initial detection model.

[0107] The above-mentioned regression coefficient loss function can be a function used to measure the difference between the model prediction value and the actual target value. The above-mentioned regression coefficient loss function can be used to guide the training process of the model, and the regression coefficient in the initial detection model can be adjusted by minimizing the function value of the loss function as much as possible to achieve better prediction performance.

[0108] In an optional embodiment, the detection system can construct a regression coefficient loss function based on the initial battery state and the sample battery state. In order to adjust the model parameters based on the loss function, the detection system can calculate the gradient of the above loss function with respect to the regression coefficient. After obtaining the gradient of the above loss function, the detection system can use the gradient descent method or its variants (such as stochastic gradient descent, gradient descent algorithm with momentum, etc.) to update the regression coefficient, so as to minimize the function value of the loss function. By continuously iterating the above adjustment process, the function value of the loss function gradually decreases. When the iteration reaches a preset number of times or the change in the loss function is lower than a preset threshold, the training process ends. The regression coefficient at this time is the above preset regression coefficient. Finally, the detection system can construct a battery state detection model based on the above preset regression coefficient and the initial detection model.

[0109] Furthermore, historical charging data of the vehicle battery on the target vehicle in the historical time period is obtained, including: obtaining the power change data and historical power of the vehicle battery in the historical time period, as well as the operating status of the vehicle battery; based on the operating status and historical power, initial charging data is filtered from the power change data; data cleaning is performed on the initial charging data to obtain cleaned charging data; missing values are filled in the cleaned charging data to obtain historical charging data.

[0110] The aforementioned operating status can be the operating conditions of the vehicle's battery at different points in time. The aforementioned initial charging data can be data within a specific time period, filtered from the vehicle's battery charge change data based on the vehicle's operating status and historical charge levels, and is primarily used for subsequent analysis and model building.

[0111] In an optional embodiment, the vehicle battery charge change data and historical charge data over historical time periods are considered, including the battery charge change over different time periods, as well as details of the vehicle's operating status, such as whether the vehicle is stationary or whether the four doors and two hoods are closed. This operating status information is particularly important for filtering out charging data collected while the vehicle is stationary, as this data is more likely to reflect the actual charging process of the battery and avoids additional interference factors that may be introduced during dynamic operation of the vehicle. Therefore, the detection system can first obtain the vehicle battery charge change data and historical charge data over historical time periods, as well as the vehicle battery's operating status. Subsequently, in order to make the power change data more stable and reliable, and to more accurately reflect the charging characteristics and health status of the battery, so as to improve the accuracy of the battery status detection model prediction, the detection system can filter out data that does not meet the conditions from the power change data based on the operating status and historical power, thereby obtaining initial charging data. However, due to measurement errors, data entry errors, etc., the above initial charging data may still contain outliers. In order to ensure that the battery status detection model is based on accurate data for learning and improve the consistency and reliability of the prediction results, the detection system can perform data cleaning on the initial charging data to remove outliers, erroneous data or irrelevant data, thereby obtaining cleaned charging data. Finally, considering that the above cleaned charging data may contain missing values due to breakpoints in the data collection process, equipment failures or unavailable data, in order to maintain data integrity, reduce data loss, and avoid deviations caused by missing data during battery status detection model training, the detection system can fill missing values in the above cleaned charging data based on interpolation of neighboring data and using global statistical values such as mean or median filling to obtain historical charging data.

[0112] For example, a detection system can collect onboard data from new energy vehicles (NEVs) using connected vehicle terminals and store it on a vehicle network big data platform for analysis and storage. To improve data calculation accuracy and reduce costs, the detection system can set the data collection frequency to 1Hz. The data collected may include at least one or more of the following: vehicle code, battery model or code, data collection time, vehicle status, high-voltage battery status, DC-DC output voltage and current, battery signals, and four-door and two-lid status, but is not limited to these. Specifically, the vehicle code uniquely identifies each vehicle. The battery model or code identifies battery characteristics, rated capacity, and other information. The data collection time records the specific time of data collection, in seconds. Vehicle status includes whether the vehicle is in motion, stopped, charging, and speed. High-voltage battery status includes battery charge, health status, and temperature, directly reflecting the condition of the NEV's high-voltage battery system. The DC-DC output voltage and current represent the output of the high-voltage battery's electrical energy through the DC-DC converter to a low-voltage voltage and current. Battery signals include battery temperature, battery current, battery voltage, and battery current accuracy. Battery current accuracy is used to evaluate the accuracy of current sensors or calibrate current measurements to ensure data accuracy. The "four doors and two hoods" feature represents the four doors, hood, and trunk lid. The "four doors and two hoods" status reflects the vehicle's safety status and passenger access.

[0113] After collecting the above data, the detection system can also pre-process the data. Specifically, the detection system can pre-process the historical data and real-time measurement data collected by the Internet of Vehicles big data platform, including data cleaning, missing value filling, outlier processing, etc., to ensure the integrity and accuracy of the data. Among them, data pre-processing includes outlier filtering and missing value interpolation. Outlier filtering is used to identify and process outliers (also known as outliers) in the data. These values may be caused by measurement errors, measurement accuracy not meeting requirements, or data entry errors. Methods for processing outliers include deleting outliers, treating outliers as missing values, or using statistical methods (such as box plots) for conversion. Missing value interpolation is used to identify and process missing values in the data. For missing values in the data, reasonable interpolation or prediction methods are used to fill them to reduce data loss. Methods for processing missing data values include using window functions to fill missing values with mean, median, mode or predicted value.

[0114] Furthermore, the method also includes: analyzing a user profile of a user who owns the target vehicle to determine the user's information acquisition characteristics, wherein the information acquisition characteristics are used to reflect the user's information acquisition habits; in response to the battery status indicating that the vehicle battery is at risk of low power, generating an information push method based on the information acquisition characteristics; and pushing risk information to the user based on the information push method, wherein the risk information is used to prompt the user that the vehicle battery is at risk of low power.

[0115] The above-mentioned information acquisition characteristics may be a collection of user preferences and habits, which affect how the user receives, digests and reacts to information.

[0116] In an optional embodiment, in order to promptly alert the user when the battery status detection model detects that the battery is low or at risk of low power, the detection system needs to be familiar with the user's information acquisition habits. Therefore, the detection system can analyze the user profile of the user who owns the target vehicle and determine the user's information acquisition characteristics. For example, the detection system can collect information on the channels used by users to interact with vehicle services through the Internet of Vehicles platform, such as the methods used by users to communicate with the service center (mobile phone app, email, text message, phone call, etc.). Subsequently, the detection system can record the time period during which users actively use Internet of Vehicles services and the days of the week when users interact more frequently with the platform. By observing the user's response to different types of information, such as the open rate, reading time, and feedback level for service updates and maintenance reminders, the user's information acquisition characteristics can be determined. Subsequently, based on the information acquisition characteristics obtained from the user profile, the detection system can customize a more appropriate information push strategy for each user group, so that when the battery status indicates that the vehicle battery is at risk of low power, the information push method can be quickly generated based on the information acquisition characteristics, thereby improving the efficiency of information push. After generating the above-mentioned information push method, the detection system can push risk information to the user based on the information push method to remind the user that the vehicle battery is at risk of low power.

[0117] For example, when the large-scale status detection model calculates that the vehicle battery is low on power or at risk of low on power, the data and information on the vehicle battery's health status will be sent to the user via the Internet, mobile phone, and telephone in the form of text, code, voice, image, and video, thereby reminding the user to conduct timely inspections and repairs to prevent the vehicle from running out of power due to battery aging.

[0118] For ease of understanding, Figure 2 This is a flow chart of vehicle battery health prediction provided by an embodiment of the present application, such as Figure 2As shown, at the beginning of the process, the detection system first collects data to obtain signals such as charging current and temperature. The collected data is then preprocessed to remove outliers and fill in missing values to ensure data integrity and consistency. After preprocessing, the detection system extracts battery recharge data segments from the historical data, further filtering out data segments where the vehicle is stationary and performing normal recharge from the high-voltage battery to the low-voltage battery. The detection system then extracts data within the same charging current interval within the recharge data segment. From this filtered recharge data segment, the detection system again sets a constant charging current range to select data segments with continuous charging current as feature data. The detection system then calculates the charge capacity and average temperature within the same recharge current interval. After completing these calculations, the detection system constructs a time series of battery decremental charging data for each vehicle, along with the mean, standard deviation, maximum, and minimum values of the feature values. This training dataset is then constructed using expert annotation of battery risk indicators, and model training is performed to develop a predictive model. After obtaining the prediction model, the detection system can input the constructed data into the prediction model to determine whether there is a power shortage or power shortage risk. If not, the process will be terminated. If so, an intelligent push will be sent to the user, and then the process will be terminated.

[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0120] Example 2

[0121] The present application also provides a battery status detection device 30, please refer to Figure 3 , including: a data acquisition module 310 for executing step S102; a fragment interception module 320 for executing step S104; a feature extraction module 330 for executing step S106, and a state detection module 340 for executing step S108.

[0122] Furthermore, the feature extraction module 330 is also used to: perform feature extraction on multiple power-replenishing data segments to obtain power-replenishing feature data of the vehicle battery, including: performing feature extraction on the power-replenishing data segments to obtain the power-replenishing time period and power-replenishing parameters corresponding to the power-replenishing data segments; determining the segment charging power and segment average temperature corresponding to the power-replenishing data segments based on the power-replenishing parameters, wherein the segment charging power is used to characterize the charging power of the vehicle battery in the corresponding power-replenishing time period, and the segment average temperature is used to characterize the average temperature of the vehicle battery in the corresponding power-replenishing time period; processing the segment charging power based on the target charging power, the power-replenishing time period, and the segment average temperature to obtain a change in the charging power of the vehicle battery in different power-replenishing time periods, wherein the target charging power is used to characterize the segment charging power corresponding to the first power-replenishing data segment among the multiple power-replenishing data segments; and constructing the power-replenishing feature data based on the charging power change and the target average temperature, wherein the target average temperature is used to characterize the segment average temperature corresponding to the first power-replenishing data segment among the multiple power-replenishing data segments.

[0123] Furthermore, the feature extraction module 330 is also used to: determine the initial charge change corresponding to different charging time periods based on the target charging power and the segment charging power; match the segment average temperature with the preset temperature range to obtain a temperature matching result, and detect the charging time period based on the preset time period to obtain a time period detection result, wherein the time detection result is used to characterize whether there are at least two charging time periods within a single preset time period; filter the initial charge change based on the temperature matching result and the time period monitoring result to obtain the charging power change.

[0124] Furthermore, the feature extraction module 330 is also used to: extract features from the charging power change to obtain the charging power features generated when the vehicle battery is recharged during a historical time period; combine the charging power features and the target average temperature according to a preset format to construct recharge feature data.

[0125] Furthermore, the status detection module 340 is also used to: perform weighted processing on the charging characteristic data based on a preset regression coefficient to obtain a weighted characteristic value; determine the probability of the vehicle battery being out of power based on the weighted characteristic value; in response to the probability of the power being out of power being greater than the preset probability, determine that the battery status is that the vehicle battery is at risk of being out of power; in response to the probability of the power being out of power being less than or equal to the preset probability, determine that the battery status is that the vehicle battery does not have a risk of being out of power.

[0126] Furthermore, the device also includes: a first acquisition module, used to obtain a sample training set of a sample battery, wherein the sample training set includes at least: sample charging data and a sample battery status of the sample battery; a first determination module, used to determine the sample charging feature data of the sample battery based on the sample charging data; a first processing module, used to process the sample charging feature data using an initial detection model to obtain an initial battery status of the vehicle battery, wherein the initial detection model includes an initial regression coefficient, and the initial regression coefficient is used to perform weighted processing on the sample charging feature data; a first adjustment module, used to adjust the initial detection model based on the initial battery status and the sample battery status to obtain a battery status detection model.

[0127] Furthermore, the first adjustment module is also used to: construct a regression coefficient loss function based on the initial battery state and the sample battery state; adjust the initial regression coefficient based on the regression coefficient loss function to obtain a preset regression coefficient; and construct a battery state detection model based on the preset regression coefficient and the initial detection model.

[0128] Furthermore, the data acquisition module 310 is also used to: obtain the power change data and historical power of the vehicle battery in a historical time period, as well as the operating status of the vehicle battery; based on the operating status and historical power, filter the initial power replenishment data from the power change data; perform data cleaning on the initial power replenishment data to obtain cleaned power replenishment data; fill in missing values in the cleaned power replenishment data to obtain historical power replenishment data.

[0129] Furthermore, the device also includes: a second determination module, used to analyze the user portrait of the user who owns the target vehicle and determine the user's information acquisition characteristics, wherein the information acquisition characteristics are used to reflect the user's information acquisition habits; a second acquisition module, used to generate an information push method based on the information acquisition characteristics in response to the battery status indicating that the vehicle battery is at risk of low power; a first push module, used to push risk information to the user based on the information push method, wherein the risk information is used to prompt the user that the vehicle battery is at risk of low power.

[0130] Example 3

[0131] The present application also provides an electronic device 40, please refer to Figure 4 , including a memory 410 and a processor 420, wherein the memory 410 is used to store computer programs; the processor 420 is used to execute the programs stored in the memory 410 to implement the battery status detection method introduced in any embodiment of the present application.

[0132] Example 4

[0133] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the battery status detection method introduced in any embodiment of the present application is implemented.

[0134] In this application, a plurality refers to two or more.

[0135] In this application, unless otherwise expressly defined, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they can refer to fixed, detachable, or integral connections; structural or electrical connections; direct or indirect connections through an intermediary; and internal communication between two components. A person of ordinary skill in the art will understand the specific meanings of these terms in this application.

[0136] The terms "first," "second," "third," "fourth," etc. (if any) in this application are used to distinguish similar objects and are not necessarily used to describe a particular sequential order.

[0137] The term "and / or" in this application simply describes an association between related objects, indicating that three possible relationships exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this application generally indicates that the related objects are in an "or" relationship.

[0138] Unless otherwise specified, all steps of the present application may be performed sequentially or randomly. For example, a statement that the method includes steps A and B indicates that the method may include steps A and B performed sequentially, or steps B and A performed sequentially. For example, a statement that the method may also include step C indicates that step C may be added to the method in any order, for example, the method may include steps A, B, and C, or steps A, C, and B, or steps C, A, and B, etc.

[0139] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A battery status detection method, characterized in that: include: Acquiring historical power replenishment data of the vehicle battery in a historical time period, wherein the historical power replenishment data is used to represent data generated when the vehicle battery is powered by a power supply battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery; Performing data analysis on the historical power replenishment data, and extracting a plurality of power replenishment data segments from the historical power replenishment data, wherein the plurality of power replenishment data segments are used to represent data segments of continuous power replenishment of the vehicle battery within a preset power replenishment current range; Performing feature extraction on the multiple power-replenishing data segments to obtain power-replenishing feature data of the vehicle battery, wherein the power-replenishing feature data is used to reflect changes in power level and temperature of the vehicle battery when the target vehicle is powered during the historical time period; The battery status detection model is used to process the charging characteristic data to obtain the battery status of the vehicle battery, wherein the battery status is used to indicate whether the vehicle battery has a risk of low power.

2. The method according to claim 1, characterized in that Extracting features from the plurality of charging data segments to obtain charging feature data of the vehicle battery includes: Performing feature extraction on the power replenishment data segment to obtain a power replenishment time period and power replenishment parameters corresponding to the power replenishment data segment; Determining, based on the power replenishment parameters, a segment charging power and a segment average temperature corresponding to the power replenishment data segment, wherein the segment charging power is used to represent the charging power of the vehicle battery during the corresponding power replenishment time period, and the segment average temperature is used to represent the average temperature of the vehicle battery during the corresponding power replenishment time period; The segment charging power is processed based on the target charging power, the charging time period, and the average temperature of the segment to obtain a charging power change of the vehicle battery in different charging time periods, wherein the target charging power is used to represent the segment charging power corresponding to the first charging data segment among the multiple charging data segments; The power replenishment characteristic data is constructed based on the charging power change and the target average temperature, wherein the target average temperature is used to represent the average temperature of the first power replenishment data segment among the multiple power replenishment data segments.

3. The method according to claim 2, characterized in that The segment charging power is processed based on the target charging power, the charging time period, and the segment average temperature to obtain a charging power change of the vehicle battery in different charging time periods, including: Determining initial power changes corresponding to different charging time periods based on the target charging power and the segmented charging power; Matching the average temperature of the segment with a preset temperature range to obtain a temperature matching result, and detecting the power-up time period based on a preset time period to obtain a time period detection result, wherein the time detection result is used to indicate whether there are at least two power-up time periods within a single preset time period; The initial power change is screened based on the temperature matching result and the time period monitoring result to obtain the charging power change.

4. The method according to claim 2, characterized in that Constructing the charging characteristic data based on the charging power change and the target average temperature includes: Extracting features of the charging power variation to obtain charging power features generated when the vehicle battery is recharged during the historical time period; The charging power characteristic and the target average temperature are combined according to a preset format to construct the charging characteristic data.

5. The method according to any one of claims 1 to 4, characterized in that Inputting the charging characteristic data into a battery status detection model, and processing the charging characteristic data using the battery status detection model to obtain the battery status of the vehicle battery, including: Performing weighted processing on the power replenishment characteristic data based on a preset regression coefficient to obtain a weighted characteristic value; determining a power-out probability of the vehicle battery based on the weighted characteristic value; In response to the battery-low probability being greater than a preset probability, determining that the battery state indicates that the vehicle battery is at risk of battery low; In response to the battery low probability being less than or equal to the preset probability, it is determined that the battery state is such that there is no risk of a battery low point for the vehicle battery.

6. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Acquire a sample training set of a sample battery, wherein the sample training set at least includes: sample charging data and a sample battery state of the sample battery; Determining sample power replenishment characteristic data of the sample battery based on the sample power replenishment data; Processing the sample charging characteristic data using the initial detection model to obtain an initial battery state of the vehicle battery, wherein the initial detection model includes an initial regression coefficient, and the initial regression coefficient is used to perform weighted processing on the sample charging characteristic data; The initial detection model is adjusted based on the initial battery state and the sample battery state to obtain the battery state detection model.

7. The method according to claim 6, characterized in that Adjusting the initial detection model based on the initial battery state and the sample battery state to obtain the battery state detection model includes: Constructing a regression coefficient loss function based on the initial battery state and the sample battery state; Adjusting the initial regression coefficient based on the regression coefficient loss function to obtain a preset regression coefficient; The battery state detection model is constructed based on the preset regression coefficient and the initial detection model.

8. The method according to any one of claims 1 to 4, characterized in that Obtain the historical charging data of the target vehicle's battery over a historical period of time, including: Obtaining power change data and historical power of the vehicle battery during the historical time period, as well as the operating status of the vehicle battery; Based on the operating status and the historical power consumption, filtering initial power replenishment data from the power consumption change data; Cleaning the initial power-replenishing data to obtain cleaned power-replenishing data; Missing values are filled in the cleaning and power-replenishing data to obtain the historical power-replenishing data.

9. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Analyzing a user profile of a user who owns the target vehicle to determine information acquisition characteristics of the user, wherein the information acquisition characteristics are used to reflect the information acquisition habits of the user; In response to the battery status indicating that the vehicle battery is at risk of being low on power, generating an information push method based on the information acquisition feature; The risk information is pushed to the user based on the information push method, wherein the risk information is used to prompt the user that the vehicle battery is at risk of low power.

10. A battery status detection device, characterized in that: include: a data acquisition module, configured to acquire historical power replenishment data of a vehicle battery on a target vehicle during a historical time period, wherein the historical power replenishment data is used to represent data generated when a power supply battery supplies power to the vehicle battery, and the voltage of the power supply battery is higher than the voltage of the vehicle battery; a fragment interception module, configured to perform data analysis on the historical power replenishment data and intercept a plurality of power replenishment data fragments from the historical power replenishment data, wherein the plurality of power replenishment data fragments are used to represent data fragments of continuous power replenishment of the vehicle battery within a preset power replenishment current range; a feature extraction module, configured to extract features from the plurality of power replenishment data segments to obtain power replenishment feature data of the vehicle battery, wherein the power replenishment feature data is configured to reflect changes in power level and temperature of the vehicle battery when power replenishment is performed on the target vehicle within the historical time period; A status detection module is used to input the charging characteristic data into a battery status detection model, and use the battery status detection model to process the charging characteristic data to obtain the battery status of the vehicle battery, wherein the battery status is used to indicate whether the vehicle battery is at risk of low power.

Images