Vehicle breakdown warning method, device, electronic device and storage medium

Abstract

The present application discloses a vehicle breakdown warning method, device, electronic device, and storage medium. The method comprises: obtaining multiple types of breakdown-related data of a target vehicle, wherein the multiple types of breakdown-related data include historical maintenance data of the target vehicle, at least one of data on the external environment in which the target vehicle is located, and vehicle operation data of the target vehicle; obtaining data features corresponding to the multiple types of breakdown-related data; performing a breakdown prediction based on the data features corresponding to the multiple types of breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to characterize whether the target vehicle has a breakdown risk. Through the above-mentioned method, the present application can improve the accuracy of vehicle breakdown prediction.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a vehicle breakdown warning method, device, electronic device and storage medium. Background Art

[0002] Currently, vehicle breakdown risk prediction is primarily based on monitoring sensor data from the vehicle's battery system (e.g., battery voltage, current, temperature, etc.). However, this prediction accuracy is low. Accurate and effective prediction of vehicle breakdown risk has become a pressing issue. Summary of the Invention

[0003] The present application at least provides a vehicle breakdown warning method, device, electronic device and storage medium.

[0004] A first aspect of the present application provides a vehicle breakdown warning method, the method comprising: obtaining a plurality of breakdown-related data of a target vehicle, wherein the plurality of breakdown-related data include historical maintenance data of the target vehicle, at least one of external environment data of the target vehicle, and vehicle operation data of the target vehicle; respectively obtaining data features corresponding to the plurality of breakdown-related data; performing a breakdown prediction based on the data features corresponding to the plurality of breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to characterize whether the target vehicle has a breakdown risk.

[0005] Therefore, the data features corresponding to the breakdown-related data can express the relationship between the breakdown-related data and the vehicle breakdown in a more concrete or explicit way. Based on the data features corresponding to the breakdown-related data, it is possible to accurately predict whether the target vehicle has a breakdown risk. In addition, the breakdown prediction is based on the data features corresponding to a variety of breakdown-related data, which can more comprehensively capture the key information that may cause the target vehicle to have a breakdown risk, thereby more accurately predicting whether the target vehicle has a breakdown risk, and then identifying the breakdown risk before the target vehicle breaks down (e.g., a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.) and issuing a breakdown risk warning, so that users can intervene in time to ensure driving safety.

[0006] Among them, the vehicle operation data includes at least one of the following: battery cell voltage data, battery cell current data, vehicle temperature data, battery pack charge status data, and vehicle operation data; the historical maintenance data includes at least one of the following: maintenance time, vehicle breakdown reason, maintenance content, vehicle status data before and after the failure, and the external environment conditions of the target vehicle when it broke down; the external environment data includes at least one of weather data and geographic data; among them, the vehicle operation data and weather data are time series data.

[0007] Therefore, the target vehicle's operating data directly reflects the target vehicle's real-time status and current health status. This data can be used to capture early signs of a breakdown, thereby predicting whether the target vehicle is likely to break down in the future. Therefore, based on the target vehicle's operating data, the target vehicle's breakdown risk can be predicted. Furthermore, vehicle operating data can be flexibly configured.

[0008] The target vehicle's historical maintenance data reveals its "health profile," revealing systemic defects, component aging trends, or hidden risk chains. This data can then be used to predict the target vehicle's potential for breakdown in the future. Consequently, the target vehicle's historical maintenance data can be used to predict its breakdown risk. Furthermore, historical maintenance data can be flexibly configured.

[0009] The target vehicle's external environment data reflects the current environmental conditions of the target vehicle. This data can be used to quantify the impact of the target vehicle's current environment on vehicle performance, identify high-risk scenarios that could lead to vehicle breakdown, and predict whether the target vehicle is likely to break down in the future. Therefore, based on the target vehicle's external environment data, the target vehicle's risk of breakdown can be predicted. Furthermore, the target vehicle's external environment data can be flexibly configured.

[0010] Among them, the cell voltage data includes the voltage value of each cell in the battery pack of the target vehicle at different times; the cell current data includes the charging and discharging current of each cell at different times; the vehicle temperature data includes at least one of the following: the cell temperature at different times, the internal ambient temperature of the target vehicle at different times; the whole vehicle operation data includes at least one of the following: the vehicle speed of the target vehicle at different times, the power output of the target vehicle at different times; the vehicle status data before and after the fault includes at least one of the following: the vehicle operation status before and after the fault, the vehicle sensor data before and after the fault; the weather data includes at least one of the following: the external ambient temperature corresponding to different times, the external ambient humidity corresponding to different times; the geographic data includes road condition information.

[0011] Therefore, the voltage of the battery cell at different times is a direct representation of the battery health status, balance and potential faults. Abnormal fluctuations, imbalances or degradation trends in the battery cell voltage can reflect the failure risk of the target vehicle's battery system. The failure of the target vehicle's battery system may cause the target vehicle to break down. Therefore, it is subsequently possible to predict whether the target vehicle has a risk of breaking down based on the voltage values of each battery cell in the target vehicle's battery pack at different times.

[0012] The charge and discharge currents of the battery cells at different times are a direct reflection of the energy flow in the target vehicle's battery system. Abnormal current patterns in the battery cells (such as overcurrent, imbalance, fluctuation, etc.) may indicate cell aging, which may cause a sudden drop in battery life and power interruption, leading to the target vehicle breaking down. Therefore, based on the charge and discharge currents of each battery cell in the target vehicle's battery pack at different times, it is possible to predict whether the target vehicle is at risk of breaking down.

[0013] Battery cell temperatures at different times are indicators of the health, safety, and reliability of the battery cells. Abnormal battery cell temperatures (e.g., localized overheating) may cause battery cell failure or a protective power outage of the target vehicle's battery system, leading to the target vehicle's breakdown. Therefore, the target vehicle's risk of breakdown can be subsequently predicted based on the battery cell temperatures of the battery pack at different times. Internal ambient temperature is a factor influencing the performance, safety, and lifespan of the target vehicle's battery system. Abnormal temperature rises, temperature drops, or uneven temperature distribution may cause battery cell failure or power limitation within the battery pack, leading to the target vehicle's breakdown. Therefore, the target vehicle's risk of breakdown can be subsequently predicted based on the internal ambient temperature of the target vehicle at different times.

[0014] The target vehicle's speed at different times is a direct reflection of the target vehicle's powertrain load, component wear, and driving behavior. Abnormal speed patterns (e.g., sustained high speeds, frequent rapid acceleration / deceleration, sudden speed drops, etc.) may indicate a powertrain failure, energy management imbalance, or hidden component failure in the target vehicle, which may cause the target vehicle to break down. Therefore, the target vehicle's speed can subsequently be used to predict whether it is at risk of breaking down. The target vehicle's power output at different times is a direct reflection of the target vehicle's powertrain health. Abnormal power fluctuations, sudden drops, or sustained over-limits may reveal potential issues such as battery cell aging, motor failure, or transmission system problems. These potential issues may lead to power outages and other problems, resulting in breakdowns. Therefore, the target vehicle's power output at different times can subsequently be used to predict whether it is at risk of breaking down.

[0015] Changes in the vehicle's operating state before and after a fault reveal the evolution of component degradation, system imbalance, or hidden failures. Therefore, the risk of a vehicle breaking down can be predicted based on the condition of its components and systems. Physical signals captured by vehicle sensors can reflect the evolution of component degradation, system imbalance, or sudden failures. Therefore, analyzing vehicle sensor data before and after a fault can predict the risk of a vehicle breaking down.

[0016] The external ambient temperature directly affects the target vehicle's material properties, chemical process stability, and system thermal management efficiency, indirectly accelerating component aging and increasing the risk of vehicle breakdown. Therefore, the target vehicle's breakdown risk can be predicted based on the corresponding external ambient temperature at different times. External humidity affects metal corrosion, electrical system reliability, sensor accuracy, and material aging, indirectly leading to component failure or system malfunction, further increasing the target vehicle's breakdown risk.

[0017] Road conditions directly affect the target vehicle's mechanical load, component wear rate, and system stability. Specific road conditions (such as bumps, slopes, and slippery conditions) can accelerate the aging of key components of the target vehicle or cause instantaneous overloads, leading to breakdowns. Therefore, the target vehicle can subsequently be predicted to be at risk of breakdown based on its road condition information.

[0018] Among them, data features corresponding to multiple anchoring-related data are obtained respectively, including: taking at least one anchoring-related data as the first anchoring-related data, and performing statistics on the first anchoring-related data to obtain data features of the first anchoring-related data; and / or, taking at least one anchoring-related data as the second anchoring-related data, and directly taking the second anchoring-related data as the data features of the second anchoring-related data.

[0019] Therefore, by converting the "implicit signal" of the target vehicle's breakdown - the first breakdown-related data, into the "explicit indicator" that can be identified by the prediction model - the data characteristics of the first breakdown-related data, the data characteristics of the first breakdown-related data can more concretely or more explicitly express the relationship between the first breakdown-related data and the vehicle breakdown. Subsequently, based on the data characteristics of the first breakdown-related data, it is possible to more accurately predict whether the target vehicle has a breakdown risk.

[0020] Since the second breakdown-related data itself can already be visualized or explicitly express the relationship between itself and the vehicle breakdown, it is directly used as its corresponding data feature.

[0021] Among them, the multiple types of breakdown-related data include vehicle operation data; in the case where the vehicle operation data includes cell voltage data, the cell voltage data is the first breakdown-related data, and the data characteristics of the cell voltage data include at least one of the following: a first cell voltage difference, a second cell voltage difference, a voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies of the battery pack within the first time window, wherein the first cell voltage difference is the voltage difference between different cells, and the second cell voltage difference is the difference between the sum of the voltages of each cell and the total voltage of the battery pack; in the case where the vehicle operation data includes cell current data, the cell current data is the first breakdown-related data, and the data characteristics of the cell current data include at least one of the following: a discharge current fluctuation characterization value corresponding to each cell, a discharge current fluctuation characterization value of the battery pack within the second The number of charging current anomalies within the time window, and the correlation between the current and voltage of each battery cell; when the vehicle operation data includes vehicle temperature data, the vehicle temperature data is the first breakdown-related data, and the data characteristics of the vehicle temperature data include at least one of the following: the temperature distribution difference of each battery cell, the temperature fluctuation characterization value of each battery cell, and the number of temperature anomalies of the battery pack within the third time window; when the vehicle operation data includes the state of charge data of the battery pack, the state of charge data is the first breakdown-related data, and the data characteristics of the state of charge data include at least one of the following: the charge fluctuation characterization value of the battery pack, the relationship between the charge consumption value of the battery pack and the mileage of the target vehicle; when the vehicle operation data includes the whole vehicle operation data, the whole vehicle operation data is the second breakdown-related data.

[0022] Therefore, the data characteristics of the cell voltage data, cell current data, vehicle temperature data, and battery pack state of charge data can be flexibly set.

[0023] Among them, the vehicle operation data is time series data, and the data features corresponding to each vehicle operation data are time series feature sequences, among which the time series feature sequences corresponding to the cell voltage data include the first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies at different moments; the time series feature sequences corresponding to the cell current data include the discharge current fluctuation characterization value, the number of charging current anomalies, and the correlation between the current and voltage of each cell at different moments; the time series feature sequences corresponding to the vehicle temperature data include the temperature distribution difference, temperature fluctuation characterization value, and the number of temperature anomalies at different moments; the time series feature sequences corresponding to the state of charge data include the temperature distribution difference, temperature fluctuation characterization value, and the number of temperature anomalies at different moments. The sequence characteristic sequence includes the relationship between the charge fluctuation characterization values corresponding to different moments, the charge value of the battery pack and the mileage of the target vehicle; and / or, the voltage fluctuation characterization value corresponding to the battery cell includes the voltage change rate of the battery cell; and / or, the first battery cell voltage difference includes the voltage difference between the maximum voltage battery cell and the minimum voltage battery cell; and / or, the discharge current fluctuation characterization value corresponding to the battery cell includes the discharge current change rate of the battery cell; and / or, the temperature distribution difference of the battery cell includes the temperature difference between the maximum temperature battery cell and the minimum temperature battery cell; and / or, the temperature fluctuation characterization value of the battery cell includes the temperature change rate of the battery cell; and / or, the charge fluctuation characterization value of the battery pack includes the charge decrease rate of the battery pack.

[0024] Therefore, the voltage fluctuation characterization value corresponding to the battery cell, the first battery cell voltage difference, the discharge current fluctuation characterization value corresponding to the battery cell, the temperature distribution difference of the battery cell, the temperature fluctuation characterization value of the battery cell and the charge fluctuation characterization value of the battery pack can be flexibly set.

[0025] Among them, when the anchoring-related data includes external environment data, and the external environment data includes external environment temperature, the external environment temperature is the first anchoring-related data, and the data characteristics of the external environment temperature include at least one of the following: the external environment temperature central trend statistical value, the external environment temperature fluctuation characterization value, and the external environment temperature distribution within the fourth time window; when the anchoring-related data includes external environment data, and the external environment data includes external environment humidity, the external environment humidity is the first anchoring-related data, and the data characteristics of the external environment humidity include at least one of the following: the external environment humidity central trend statistical value, the external environment humidity fluctuation characterization value; when the anchoring-related data includes historical maintenance data of the target vehicle, the historical maintenance data is the first anchoring-related data, and the data characteristics of the historical maintenance data include at least one of the following: the total number of historical anchoring times of the target vehicle, the number of times the target vehicle has been anchored due to cell failure, the statistical results of parts replacement of the target vehicle, the fault code of the fault that caused the target vehicle to be anchored, the anchoring interval of the target vehicle, the vehicle operation data of the target vehicle before the anchoring, and the vehicle operation status of the target vehicle after the anchoring repair.

[0026] Therefore, the data characteristics of the external environment temperature, external environment humidity, and historical maintenance data can be flexibly set.

[0027] Among them, the external ambient temperature and the external ambient humidity are time series data, and the data features corresponding to the external ambient temperature and the external ambient humidity are time series feature sequences, wherein the time series feature sequence corresponding to the external ambient temperature includes the external ambient temperature central trend statistical values corresponding to different moments, the external ambient temperature fluctuation characterization values, and the external ambient temperature distribution; the time series feature sequence corresponding to the external ambient humidity includes the external ambient humidity central trend statistical values corresponding to different moments, the external ambient humidity fluctuation characterization values; and / or, the external ambient temperature fluctuation characterization values include the external ambient temperature change rate; and / or, the external ambient temperature distribution includes the highest temperature and the lowest temperature within the fourth time window; and / or, the external ambient humidity fluctuation characterization values include the external ambient humidity change rate.

[0028] Therefore, the external environment temperature fluctuation characterization value, the external environment temperature distribution and the external environment humidity fluctuation characterization value can be flexibly set.

[0029] Among them, before performing a breakdown prediction based on the data features corresponding to a plurality of breakdown-related data and obtaining the breakdown prediction result, the vehicle breakdown warning method also includes at least one of the following steps: finding an associated feature group whose feature correlation meets a preset correlation condition from the data features corresponding to the plurality of breakdown-related data, selecting a representative feature from the associated feature group, and deleting other data features in the associated feature group except the representative feature; finding each time series feature sequence from the data features corresponding to the plurality of breakdown-related data, and combining each time series feature sequence into a time series data feature.

[0030] Therefore, the feature correlations between the data features in the associated feature group meet the preset correlation conditions, indicating that the data features in the associated feature group are highly correlated and may contain information overlap. Therefore, representative features are selected from the associated feature group and retained, while other data features in the associated feature group are deleted. This effectively reduces redundant features in the data features corresponding to the various anchor correlation data, while retaining key features in the data features corresponding to the various anchor correlation data. In other words, redundant features are identified and deleted by quantifying the feature correlations between the data features corresponding to the various anchor correlation data.

[0031] Combining the time series feature sequences can enhance the ability to capture hidden relationships in time series data, thereby improving the accuracy of subsequent vehicle breakdown prediction based on time series data features.

[0032] Among them, the data features corresponding to the various anchoring-related data include time series data features and non-time series data features; anchoring prediction is performed based on the data features corresponding to the various anchoring-related data to obtain anchoring prediction results, including: using a time series processing model to detect the time series dependency of time series data features to obtain time-dependent features; and using a decision model to identify the association between non-time series data features and anchoring to obtain anchoring-related features; using time-dependent features and anchoring-related features to perform prediction to obtain anchoring prediction results.

[0033] Therefore, time-dependent features are derived by using a time series processing model to detect the temporal dependencies of time series data features. They can reflect the long-term decay trends (e.g., battery capacity decline) and short-term abnormal fluctuations (e.g., battery temperature surges) of the target vehicle's time series data, modeling its dynamic evolution. Breakdown-related features are derived by using a decision model to identify the association between non-time series data features and breakdowns, enabling exploration of the correlations between discrete events. Therefore, using both time-dependent and breakdown-related features for prediction simultaneously covers both "gradual aging" and "sudden impact" risks, avoiding the blind spots of a single data source and improving the accuracy of vehicle breakdown risk prediction.

[0034] Among them, the time series processing model is a long short-term memory network model, and the decision model is a random forest model; and / or, time-dependent features and anchoring correlation features are used to make predictions to obtain anchoring prediction results, including: fusing time-dependent features and anchoring correlation features to obtain fused features; and making predictions based on the fused features to obtain anchoring prediction results.

[0035] Therefore, the timing processing model and decision model can be set flexibly.

[0036] By fusing time-dependent features with anchoring-related features, a multi-dimensional and multi-granular anchoring risk characterization system is constructed, thereby improving the accuracy of anchoring risk prediction.

[0037] Among them, multiple types of breakdown-related data include time series data and non-time series data; before respectively obtaining the data features corresponding to the multiple types of breakdown-related data, the vehicle breakdown warning method also includes: performing a first preprocessing on the time series data; wherein, the first preprocessing includes at least one of the following: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data alignment; and, performing a second preprocessing on the non-time series data; wherein, the second preprocessing includes at least one of the following: data format unification processing, missing data completion, and data alignment.

[0038] Therefore, a first preprocessing is performed on the time series data among the multiple anchor-related data, and a second preprocessing is performed on the non-time series data among the multiple anchor-related data, so as to convert the time series data and non-time series data among the multiple anchor-related data into high-quality data. Subsequently, anchor prediction is performed based on the preprocessed multiple anchor-related data, which can improve the accuracy of anchor prediction.

[0039] Among them, the breakdown-related data includes the vehicle operation data of the target vehicle, and the vehicle operation data is time series data; the first preprocessing of the time series data includes: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data time alignment of the vehicle operation data respectively; and / or, the breakdown-related data includes the historical maintenance data of the target vehicle, and the historical maintenance data is non-time series data; the second preprocessing of the non-time series data includes: data format unification processing and missing data completion of the historical maintenance data of the target vehicle respectively; and / or, the breakdown-related data includes the vehicle operation data of the target vehicle and the external environment data of the target vehicle, and the external environment data includes weather data and geographic data, the vehicle operation data and weather data are time series data, and the geographic data is non-time series data; the step of preprocessing the breakdown-related data also includes: data time alignment of the vehicle operation data and the weather data; and data space alignment of the vehicle operation data and the geographic data.

[0040] Therefore, the pre-processing of the vehicle operation data of the target vehicle, the historical maintenance data of the target vehicle and the external environment data of the target vehicle can be flexibly set.

[0041] Among them, the first prediction result is used to characterize the risk of breakdown of the target vehicle, and the breakdown prediction result also includes a second prediction result, and the second prediction result includes at least one of the following: the predicted occurrence time of the breakdown risk, the predicted cause of the breakdown risk, and the solution to the breakdown risk; and / or, after performing breakdown prediction based on data features corresponding to multiple breakdown-related data and obtaining the breakdown prediction result, the vehicle breakdown warning method also includes: feeding back at least part of the information in the breakdown prediction result to the user, wherein the feedback method includes at least one of the following: displaying at least part of the information on the display screen of the target vehicle, playing at least part of the information through audio on the target vehicle, and sending at least part of the information to an associated terminal.

[0042] Therefore, when performing breakdown prediction based on the data features corresponding to a variety of breakdown-related data and determining that the target vehicle is at risk of breakdown, relevant information such as the time of occurrence of the breakdown risk, the cause of the breakdown risk, and the solution measures for the breakdown risk will also be predicted; that is, the breakdown prediction results are diverse and rich.

[0043] There are various ways to feed back at least part of the information in the anchor prediction result to the user, and the feedback method can be flexibly selected.

[0044] A second aspect of the present application provides a vehicle breakdown warning device, which includes a first acquisition module, a second acquisition module and a prediction module; the first acquisition module is used to acquire multiple breakdown-related data of a target vehicle, wherein the multiple breakdown-related data include historical maintenance data of the target vehicle, at least one of the external environment data of the target vehicle, and vehicle operation data of the target vehicle; the second acquisition module is used to respectively acquire data features corresponding to the multiple breakdown-related data; the prediction module is used to perform breakdown prediction based on the data features corresponding to the multiple breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to characterize whether the target vehicle has a breakdown risk.

[0045] A third aspect of the present application provides an electronic device, which includes a memory and a processor, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the above-mentioned vehicle breakdown warning method.

[0046] A fourth aspect of the present application provides a computer-readable storage medium for storing program instructions, which can be executed to implement the above-mentioned vehicle breakdown warning method.

[0047] In the above technical solution, the data features corresponding to the breakdown-related data can express the relationship between the breakdown-related data and the vehicle breakdown in a more concrete or explicit way. Based on the data features corresponding to the breakdown-related data, it is possible to accurately predict whether the target vehicle has a breakdown risk. In addition, the breakdown prediction is performed based on the data features corresponding to multiple types of breakdown-related data, which can more comprehensively capture key information that may cause the target vehicle to have a breakdown risk, thereby more accurately predicting whether the target vehicle has a breakdown risk, and further, it can identify the breakdown risk and issue a breakdown risk warning before the target vehicle breaks down (e.g., a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.), so that the user can intervene in time to ensure driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of an embodiment of a vehicle breakdown warning method provided by the present application;

[0049] Figure 2 yes Figure 1 The flowchart of step S13 is shown as an embodiment;

[0050] Figure 3 This is a structural diagram of an embodiment of a vehicle breakdown warning device provided by the present application;

[0051] Figure 4 This is a structural diagram of an embodiment of an electronic device provided by the present application;

[0052] Figure 5 It is a structural diagram of an embodiment of a computer-readable storage medium provided by this application. DETAILED DESCRIPTION

[0053] The following describes the embodiments of the present application in detail with reference to the accompanying drawings.

[0054] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0055] The term "and / or" in this article is simply a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects are in an "or" relationship. In addition, "many" in this article means two or more than two. In addition, the term "at least one" in this article means any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0056] See also Figure 1 , Figure 1 This is a flow chart of an embodiment of the vehicle breakdown warning method provided by this application. It should be noted that if there are substantially the same results, this embodiment is not based on Figure 1 The process sequence shown is limited. Figure 1 As shown, this embodiment includes:

[0057] Step S11: Acquire various breakdown-related data of the target vehicle.

[0058] In this embodiment, a variety of breakdown-related data of the target vehicle is obtained; wherein, the various breakdown-related data include the historical maintenance data of the target vehicle, at least one of the external environment data of the target vehicle, and the vehicle operation data of the target vehicle. The vehicle operation data of the target vehicle directly reflects the real-time status and current health status of the target vehicle. The vehicle operation data of the target vehicle can capture early signals of vehicle breakdown, thereby predicting whether the target vehicle is likely to break down in the future; therefore, it is possible to subsequently predict whether the target vehicle is at risk of breakdown based on the vehicle operation data of the target vehicle. The historical maintenance data of the target vehicle reveals the "health profile" of the target vehicle. The historical maintenance data of the target vehicle can reveal the target vehicle's systemic defects, component aging trends, or hidden risk chains, thereby predicting whether the target vehicle is likely to break down in the future; therefore, it is possible to subsequently predict whether the target vehicle is at risk of breakdown based on the historical maintenance data of the target vehicle. The external environment data of the target vehicle reflects the environmental conditions of the target vehicle's current environment. Through the external environment data of the target vehicle, the impact of the target vehicle's current environment on the vehicle performance can be quantified, and high-risk scenarios that may cause the vehicle to break down can be identified, thereby predicting whether the target vehicle may break down in the future; therefore, it can be predicted whether the target vehicle is at risk of breaking down based on the external environment data of the target vehicle.

[0059] There is no limitation on the types of the various breakdown-related data obtained for the target vehicle, and these types can be specifically set according to actual use needs.

[0060] For example, the multiple breakdown-related data of the target vehicle obtained include vehicle operation data of the target vehicle and historical maintenance data of the target vehicle. For another example, the multiple breakdown-related data of the target vehicle obtained include vehicle operation data of the target vehicle and external environment data of the target vehicle.

[0061] For another example, the various types of breakdown-related data acquired for a target vehicle include the target vehicle's operating data, historical maintenance data, and data about the target vehicle's external environment. When these data are combined, a panoramic portrait of the target vehicle is constructed from the three dimensions of the target vehicle's real-time status, health history, and environmental adaptability. This allows for more comprehensive capture of subtle signs of a breakdown, or in other words, more comprehensive capture of key information that may indicate a breakdown risk. Furthermore, subsequent breakdown prediction based on these data can more accurately predict whether the target vehicle is at risk of breakdown. This allows for identification of breakdown risks and issuance of a breakdown risk warning before a breakdown occurs (e.g., hours or days before a breakdown occurs), enabling timely user intervention to ensure driving safety.

[0062] In one embodiment, the vehicle operation data includes at least one of the following: cell voltage data, cell current data, vehicle temperature data, battery pack state of charge data, and vehicle operation data. Cell voltage data, cell current data, vehicle temperature data, battery pack state of charge data, and vehicle operation data are essentially the "vital signs" of the target vehicle's powertrain, directly reflecting the target vehicle's real-time status and current health. By using these data, early signs of vehicle breakdown can be captured, thereby predicting the likelihood of the target vehicle breaking down in the future.

[0063] In a specific embodiment, the cell voltage data includes the voltage value of each cell in the battery pack of the target vehicle at different times. The voltage of the cell at different times is a direct representation of the battery health status, balance and potential faults. The abnormal fluctuation, imbalance or degradation trend of the cell voltage can reflect the failure risk of the battery system of the target vehicle. The failure of the battery system of the target vehicle may cause the target vehicle to break down. For example, a sudden drop or imbalance in the cell voltage may trigger a protective power outage of the BMS, causing the vehicle to break down instantly; for another example, the degradation of the cell voltage indicates capacity attenuation, and a cliff-like drop in cruising range may cause the vehicle to break down midway. Therefore, it is subsequently possible to predict whether the target vehicle is at risk of breaking down based on the voltage value of each cell in the battery pack of the target vehicle at different times.

[0064] In one specific embodiment, the cell current data includes the charge and discharge currents of each cell at different times. The charge and discharge currents of the cells at different times directly reflect the energy flow of the target vehicle's battery system. Abnormal current patterns in the cells (e.g., overcurrent, imbalance, fluctuation, etc.) may indicate cell aging, which can cause a sudden drop in range, power outages, and even the target vehicle to break down. Therefore, based on the charge and discharge currents of each cell in the target vehicle's battery pack at different times, it is possible to predict whether the target vehicle is at risk of breaking down.

[0065] In one specific embodiment, the vehicle temperature data includes at least one of the following: battery cell temperatures at different times, and the internal ambient temperature of the target vehicle at different times. The battery cell temperatures at different times are indicators of the health, safety, and reliability of the battery cells. Abnormal battery cell temperatures (e.g., localized overheating) may cause battery cell failure or a protective power outage of the target vehicle's battery system, leading to the target vehicle breaking down. Therefore, the target vehicle can subsequently be predicted to be at risk of breaking down based on the battery cell temperatures of the target vehicle's battery pack at different times. The internal ambient temperature of the target vehicle at different times can be considered the internal ambient temperature of the battery pack. The internal ambient temperature is a factor that affects the performance, safety, and lifespan of the target vehicle's battery system. Abnormal temperature rises, temperature drops, or uneven temperature distribution may cause battery cell failure or power limitation within the battery pack, leading to the target vehicle breaking down. Therefore, the target vehicle can subsequently be predicted to be at risk of breaking down based on the internal ambient temperature of the target vehicle at different times.

[0066] In one specific embodiment, the vehicle operation data includes at least one of the following: the target vehicle's speed at different times, and the target vehicle's power output at different times. The target vehicle's speed at different times is a direct reflection of the target vehicle's powertrain load, component wear, and driving behavior. Abnormal speed patterns (e.g., sustained high speeds, frequent rapid acceleration / deceleration, sudden speed drops, etc.) may indicate a powertrain failure, energy management imbalance, or hidden component failure in the target vehicle, which may cause the target vehicle to break down. The target vehicle's power output at different times is a direct reflection of the target vehicle's powertrain health. Abnormal power fluctuations, sudden drops, or sustained over-limit may reveal potential issues such as battery cell aging, motor failure, or transmission system problems. These potential issues may lead to power outages and breakdowns.

[0067] In a specific implementation, the vehicle operation data of the target vehicle may be obtained from an on-board diagnostics system (OBD) or a battery management system (BMS).

[0068] In one specific embodiment, the target vehicle's operating data in the onboard diagnostic system or battery management system may be detected by sensors installed on the target vehicle. For example, the target vehicle's operating data may include the voltage value of each cell in the target vehicle's battery pack at different times, the charge and discharge current of each cell at different times, the cell temperature of each cell at different times, the internal ambient temperature of the target vehicle at different times, the vehicle speed of the target vehicle at different times, the power output of the target vehicle at different times, the acceleration of the target vehicle at different times, the vibration frequency of the target vehicle at different times, and the state of charge data of the target vehicle's battery pack. Among them, the voltage value of each cell in the battery pack of the target vehicle at different times is obtained by detecting the voltage sensor installed on the target vehicle; the charging and discharging current of each cell in the battery pack of the target vehicle at different times is obtained by detecting the current sensor installed on the target vehicle; the cell temperature of each cell in the battery pack of the target vehicle at different times is obtained by detecting the first temperature sensor installed on the target vehicle; the internal ambient temperature of the target vehicle at different times is obtained by detecting the second temperature sensor installed on the target vehicle; the vibration frequency of the target vehicle at different times is obtained by detecting the vibration sensor installed on the target vehicle; the acceleration of the target vehicle at different times is obtained by detecting the acceleration sensor installed on the target vehicle; the vehicle driving speed of the target vehicle at different times is obtained by detecting the vehicle speed sensor installed on the target vehicle; the power output of the target vehicle at different times is obtained by detecting the torque sensor installed on the target vehicle; the charge state data of the battery pack of the target vehicle is obtained by detecting the current sensor installed on the target vehicle.

[0069] In one embodiment, historical maintenance data includes at least one of the following: maintenance time, breakdown cause (e.g., battery failure, sensor failure, software issue, etc.), maintenance details (e.g., replaced parts, repair measures, software updates, etc.), vehicle status data before and after the breakdown, and the external environment conditions of the target vehicle at the time of breakdown. This historical maintenance data, including the target vehicle's maintenance time, breakdown cause, maintenance details, vehicle status data before and after the breakdown, and the external environment conditions at the time of breakdown, constitutes the target vehicle's "health profile." This "health profile" reveals systemic defects, component aging trends, or hidden risk chains. Therefore, based on this historical maintenance data, including the target vehicle's maintenance time, breakdown cause, maintenance details, vehicle status data before and after the breakdown, and the external environment conditions at the time of breakdown, it is possible to predict whether the target vehicle is at risk of breakdown.

[0070] Of course, in other implementations, the historical maintenance data may also include maintenance frequency, maintenance cycle, etc., which are not limited here.

[0071] In one specific embodiment, the vehicle status data before and after a fault includes at least one of the following: the vehicle's operating status before and after the fault, and vehicle sensor data before and after the fault. Regarding the vehicle's operating status before and after the fault, changes in the vehicle's operating status before and after the fault reveal the evolution of component degradation, system imbalance, or latent failures in the target vehicle. Therefore, the presence of a target vehicle at risk of breakdown can be predicted based on the conditions of the target vehicle's components and systems. Regarding the vehicle sensor data before and after a fault, the physical signals captured by the vehicle sensors can reflect the evolution of component degradation, system imbalance, or sudden failures in the target vehicle. Therefore, analyzing the vehicle sensor data before and after a fault can predict the presence of a target vehicle at risk of breakdown.

[0072] In a specific embodiment, the historical maintenance data of the target vehicle can be obtained from the after-sales service system of the target vehicle, the maintenance report of the target vehicle, the fault diagnosis system (such as OBD), the vehicle owner, etc.

[0073] In one embodiment, the external environment data includes at least one of weather data and geographic data. Weather data is a significant external variable that impacts the operational reliability of the target vehicle. Extreme or exceptional weather conditions can exacerbate component wear, material aging, or system performance degradation, thereby increasing the target vehicle's risk of breakdown. Therefore, the target vehicle's weather data can subsequently be used to predict whether the target vehicle is at risk of breakdown. Geographic data, the geographic environment, by affecting the physical loads on target vehicle components, material aging rates, and system adaptability, can directly or indirectly lead to mechanical wear, thermal management failure, or energy demand imbalance, thereby increasing the risk of breakdown. Therefore, the target vehicle's geographic data can subsequently be used to predict whether the target vehicle is at risk of breakdown.

[0074] In one specific embodiment, the weather data includes at least one of the following: external ambient temperature at different times and external ambient humidity at different times. The external ambient temperature at different times directly affects the material properties, chemical process stability, and system thermal management efficiency of the target vehicle, thereby indirectly accelerating component aging and increasing the target vehicle's risk of breakdown. Therefore, the risk of breakdown can be subsequently predicted based on the external ambient temperature at different times. The external ambient humidity at different times can indirectly cause component failure or system malfunction by affecting metal corrosion, electrical system reliability, sensor accuracy, and material aging, thereby increasing the target vehicle's risk of breakdown.

[0075] Of course, in other specific implementations, the weather data also includes whether there is rain in the external environment where the target vehicle is located, whether there is snow in the external environment where the target vehicle is located, the wind speed of the external environment corresponding to different times, etc., which are not limited here and can be specifically set according to actual use needs.

[0076] In one specific embodiment, the geographic data includes road condition information. Road conditions directly impact the target vehicle's mechanical load, component wear rate, and system stability. Certain road conditions (e.g., bumpy, sloped, slippery, etc.) can accelerate the aging of key components or cause transient overload, leading to breakdowns. Therefore, based on the target vehicle's road condition information, it is possible to subsequently predict whether the target vehicle is at risk of breakdown.

[0077] The road condition information may include whether the road is slippery, whether the road has a slope, whether the road has a curve, etc., which are not limited here.

[0078] In a specific implementation, the external environment data includes weather data, and specifically, the weather data of the target vehicle can be obtained from a weather API service.

[0079] In a specific embodiment, the external environment data includes geographic data. Specifically, the vehicle position of the target vehicle can be obtained through the GPS of the target vehicle, and then the vehicle position of the target vehicle is matched with the high-precision map to determine the geographic data of the vehicle position of the target vehicle.

[0080] It should be noted that the vehicle operation data and weather data of the target vehicle are time series data. Time series data refers to a series of data points recorded in chronological order. Each data point has a timestamp (Timestamp) that is used to represent the state or measurement value of the data at a certain point in time. The characteristic of time series data is that there is a temporal dependency between the data, that is, the current data may be affected by past data, and the data may show trends, periodicity or random fluctuations over time.

[0081] Non-time series data refers to data that is not recorded in a chronological order. There is no strict temporal relationship between data points, or the time dimension is not a primary consideration in the analysis. This type of data typically reflects static attributes, discrete events, or fixed states, rather than dynamic processes that change over time.

[0082] Step S12: respectively obtaining data features corresponding to a plurality of anchoring-related data.

[0083] In this embodiment, data features corresponding to various types of breakdown-related data are separately acquired. In other words, the "implicit signals" of a target vehicle's breakdown—the target vehicle's various breakdown-related data—are converted into "explicit indicators" recognizable by the prediction model—the data features corresponding to the various types of breakdown-related data. These data features can more concretely or explicitly express the relationship between the breakdown-related data and the vehicle's breakdown. Subsequently, based on the data features corresponding to the various types of breakdown-related data, a more accurate prediction of whether the target vehicle is at risk of breakdown can be made. This allows the target vehicle to identify breakdown risks before they occur (e.g., hours or days before a breakdown occurs) and issue a breakdown risk warning, enabling timely user intervention to ensure driving safety.

[0084] In one embodiment, data features corresponding to multiple types of breakdown-related data are obtained separately. Specifically, at least one type of breakdown-related data is taken as first breakdown-related data, and statistics are collected on the first breakdown-related data to obtain data features of the first breakdown-related data. In other words, statistics are collected on some of the breakdown-related data, and the statistical results are used as corresponding data features. By converting the first breakdown-related data—the "implicit signal" of a target vehicle's breakdown—into "explicit indicators" recognizable by the prediction model—the data features of the first breakdown-related data—the data features of the first breakdown-related data can more concretely or explicitly express the relationship between the first breakdown-related data and vehicle breakdown. Subsequently, based on the data features of the first breakdown-related data, a more accurate prediction of whether the target vehicle is at risk of breakdown can be made. This allows the breakdown risk to be identified and issued before the target vehicle breaks down (e.g., hours or days before a breakdown occurs), enabling timely user intervention and ensuring driving safety.

[0085] In a specific embodiment, the multiple breakdown-related data include vehicle operation data; when the vehicle operation data includes battery cell voltage data, the battery cell voltage data is the first breakdown-related data, and the data characteristics of the battery cell voltage data include at least one of the following: a first battery cell voltage difference, a second battery cell voltage difference, a voltage fluctuation characterization value corresponding to each battery cell, and the number of voltage anomalies of the battery pack within a first time window, wherein the first battery cell voltage difference is the voltage difference between different battery cells, and the second battery cell voltage difference is the difference between the sum of the voltages of each battery cell and the total voltage of the battery pack.

[0086] The "implicit signal" of the target vehicle's breakdown - the cell voltage data, is converted into an "explicit indicator" that can be identified by the prediction model - the first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies of the battery pack within the first time window. The first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies of the battery pack within the first time window can more concretely or explicitly express the relationship between the cell voltage data and the vehicle breakdown. Subsequently, based on the first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies of the battery pack within the first time window, it is possible to more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (e.g., several hours before the breakdown occurs, several days before the breakdown occurs, etc.), so that the user can intervene in time to ensure driving safety.

[0087] The size of the first time window is not limited and can be set according to actual usage needs. In addition, it should be noted that the abnormal voltage of the battery pack within the first time window can be understood as a voltage that is too low or too high, such as a voltage that is higher than a first threshold or lower than a second threshold.

[0088] In a specific embodiment, the voltage fluctuation characterization value corresponding to the battery cell includes the voltage change rate of the battery cell.

[0089] In one embodiment, the first cell voltage difference includes the voltage difference between the cell with the maximum voltage and the cell with the minimum voltage. Of course, in other embodiments, the first cell voltage difference may also be the voltage difference between any two cells, which is not limited here.

[0090] In a specific embodiment, the multiple breakdown-related data include vehicle operation data; when the vehicle operation data includes battery cell current data, the battery cell current data is the first breakdown-related data, and the data characteristics of the battery cell current data include at least one of the following: the discharge current fluctuation characterization value corresponding to each battery cell, the number of abnormal charging currents of the battery pack within the second time window, and the correlation between the current and voltage of each battery cell.

[0091] Converting the "implicit signal" of the target vehicle's breakdown - the cell current data - into an "explicit indicator" that can be identified by the prediction model - the discharge current fluctuation characterization value corresponding to each cell, the number of abnormal charging current of the battery pack in the second time window, and the correlation between the current and voltage of each cell. This can more concretely or explicitly express the relationship between the cell current data and the vehicle breakdown. Subsequently, based on the discharge current fluctuation characterization value corresponding to each cell, the number of abnormal charging current of the battery pack in the second time window, and the correlation between the current and voltage of each cell, it can more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (e.g., a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.), so that users can intervene in time to ensure driving safety.

[0092] The size of the second time window is not limited and can be set according to actual usage needs. In addition, it should be noted that the abnormal charging current of the battery pack within the second time window can be understood as the charging current being too low or too high, such as the charging current being higher than the third threshold or lower than the fourth threshold. Furthermore, the correlation between the current and voltage of each battery cell can be represented by whether they are positively correlated or negatively correlated, as well as the corresponding correlation coefficient.

[0093] In a specific embodiment, the discharge current fluctuation characterization value corresponding to the battery cell includes the discharge current change rate of the battery cell.

[0094] In a specific embodiment, the multiple breakdown-related data include vehicle operation data; when the vehicle operation data includes vehicle temperature data, the vehicle temperature data is the first breakdown-related data, and the data characteristics of the vehicle temperature data include at least one of the following: the temperature distribution difference of each battery cell, the temperature fluctuation characterization value of each battery cell, and the number of temperature anomalies of the battery pack within the third time window.

[0095] Converting the "implicit signal" of the target vehicle's breakdown - vehicle temperature data - into "explicit indicators" that can be identified by the prediction model - the temperature distribution differences of each battery cell, the temperature fluctuation characterization value of each battery cell, and the number of temperature anomalies of the battery pack within the third time window. This can more concretely or explicitly express the relationship between vehicle temperature data and vehicle breakdown. Subsequently, based on the temperature distribution differences of each battery cell, the temperature fluctuation characterization value of each battery cell, and the number of temperature anomalies of the battery pack within the third time window, it can more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (e.g., a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.), so that users can intervene in time to ensure driving safety.

[0096] The size of the third time window is not limited and can be set according to actual usage needs. In addition, it should be noted that an abnormal battery pack temperature within the third time window can be understood as the battery pack temperature being too low or too high, for example, the temperature being higher than the fifth threshold or lower than the sixth threshold.

[0097] In one embodiment, the temperature distribution difference of the battery cells includes the temperature difference between the battery cell with the maximum temperature and the battery cell with the minimum temperature. Of course, in other embodiments, the temperature distribution difference of the battery cells may also be the temperature difference between any two battery cells, which is not limited here.

[0098] In a specific embodiment, the temperature fluctuation characterization value of the battery cell includes the temperature change rate of the battery cell.

[0099] In a specific embodiment, the multiple breakdown-related data include vehicle operation data; when the vehicle operation data includes the charge state data of the battery pack, the charge state data is the first breakdown-related data, and the data characteristics of the charge state data include at least one of the following: the charge fluctuation characterization value of the battery pack, the relationship between the charge consumption value of the battery pack and the mileage of the target vehicle.

[0100] Converting the "implicit signal" of the target vehicle's breakdown - the battery pack's state of charge data - into an "explicit indicator" that can be identified by the prediction model - the relationship between the battery pack's charge fluctuation characterization value and the battery pack's charge consumption value and the target vehicle's mileage. This can more concretely or explicitly express the relationship between the battery pack's state of charge data and the vehicle's breakdown. Subsequently, based on the relationship between the battery pack's charge fluctuation characterization value and the battery pack's charge consumption value and the target vehicle's mileage, it can more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (e.g., several hours before the breakdown occurs, several days before the breakdown occurs, etc.), so that users can intervene in time to ensure driving safety.

[0101] In one embodiment, the charge fluctuation characteristic value of the battery pack includes a charge decrease rate of the battery pack.

[0102] In a specific embodiment, the vehicle operation data is time series data, and the data features corresponding to each vehicle operation data are time series feature sequences, wherein the time series feature sequence corresponding to the battery cell voltage data includes the first battery cell voltage difference, the second battery cell voltage difference, the voltage fluctuation characterization value corresponding to each battery cell, and the number of voltage anomalies corresponding to different moments; the time series feature sequence corresponding to the battery cell current data includes the discharge current fluctuation characterization value corresponding to different moments, the number of charging current anomalies, and the correlation between the current and voltage of each battery cell; the time series feature sequence corresponding to the vehicle temperature data includes the temperature distribution difference, the temperature fluctuation characterization value, and the number of temperature anomalies corresponding to different moments; the time series feature sequence corresponding to the charge state data includes the charge fluctuation characterization value corresponding to different moments, and the relationship between the charge value of the battery pack and the mileage of the target vehicle.

[0103] In a specific embodiment, when the anchoring-related data includes external environment data, and the external environment data includes external environment temperature, the external environment temperature is the first anchoring-related data, and the data characteristics of the external environment temperature include at least one of the following: a statistical value of the central trend of the external environment temperature, a characterization value of the fluctuation of the external environment temperature, and the distribution of the external environment temperature within the fourth time window.

[0104] Converting the "implicit signal" of the target vehicle's breakdown - the external ambient temperature - into "explicit indicators" that can be identified by the prediction model - the external ambient temperature central trend statistical value, the external ambient temperature fluctuation characterization value, and the external ambient temperature distribution within the fourth time window, can more concretely or explicitly express the relationship between the external ambient temperature and vehicle breakdown. Subsequently, based on the external ambient temperature central trend statistical value, the external ambient temperature fluctuation characterization value, and the external ambient temperature distribution within the fourth time window, it can more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (e.g., a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.), so that users can intervene in time to ensure driving safety.

[0105] There is no limitation on the size of the fourth time window, and it can be set according to actual needs.

[0106] In a specific embodiment, the external environment temperature fluctuation representative value includes the external environment temperature change rate.

[0107] In a specific embodiment, the external environment temperature distribution includes a maximum temperature and a minimum temperature within the fourth time window.

[0108] In one embodiment, the external environment temperature central tendency statistical value may be the average value of the external environment temperature, that is, the external average environment temperature. Of course, in other embodiments, the external environment temperature central tendency statistical value may also be the median value of the external environment temperature, which is not limited here.

[0109] In a specific embodiment, when the anchoring-related data includes external environment data, and the external environment data includes external environment humidity, the external environment humidity is the first anchoring-related data, and the data characteristics of the external environment humidity include at least one of the following: a statistical value of the central trend of the external environment humidity, and a characterization value of the fluctuation of the external environment humidity.

[0110] Converting the "implicit signal" of the target vehicle's breakdown - the external environmental humidity - into an "explicit indicator" that can be identified by the prediction model - the external environmental humidity central trend statistical value and the external environmental humidity fluctuation characterization value, can more concretely or more explicitly express the relationship between the external environmental humidity and vehicle breakdown. Subsequently, based on the external environmental humidity central trend statistical value and the external environmental humidity fluctuation characterization value, it can more accurately predict whether the target vehicle has a breakdown risk, thereby identifying the breakdown risk and issuing a breakdown risk warning before the target vehicle breaks down (for example, a few hours before the breakdown occurs, a few days before the breakdown occurs, etc.), so that users can intervene in time to ensure driving safety.

[0111] In a specific embodiment, the external environment humidity fluctuation representative value includes the external environment humidity change rate.

[0112] In one embodiment, the external environment humidity central tendency statistical value may be the average value of the external environment humidity, that is, the external average environment humidity. Of course, in other embodiments, the external environment humidity central tendency statistical value may also be the median value of the external environment humidity, which is not limited here.

[0113] In a specific embodiment, the external ambient temperature and the external ambient humidity are time series data, and the data features corresponding to the external ambient temperature and the external ambient humidity are time series feature sequences, wherein the time series feature sequences corresponding to the external ambient temperature include the external ambient temperature central trend statistical values corresponding to different moments, the external ambient temperature fluctuation characterization values, and the external ambient temperature distribution; the time series feature sequences corresponding to the external ambient humidity include the external ambient humidity central trend statistical values corresponding to different moments, the external ambient temperature fluctuation characterization values.

[0114] In a specific embodiment, when the breakdown-related data includes historical maintenance data of the target vehicle, the historical maintenance data is the first breakdown-related data, and the data characteristics of the historical maintenance data include at least one of the following: the total number of historical breakdowns of the target vehicle, the number of times the target vehicle has broken down due to battery cell failure, the statistical results of parts replacement of the target vehicle, the fault code of the fault that caused the target vehicle to break down, the breakdown interval of the target vehicle, the vehicle operation data of the target vehicle before the breakdown, and the vehicle operation status of the target vehicle after the breakdown repair.

[0115] The "implicit signal" of the target vehicle's breakdown—historical maintenance data—is converted into "explicit indicators" that can be identified by the prediction model. These indicators include the target vehicle's total historical breakdown count, the number of target vehicle breakdowns caused by battery cell failures, the target vehicle's parts replacement statistics, the fault codes of the faults that caused the target vehicle's breakdowns, the target vehicle's breakdown intervals, the target vehicle's vehicle operating data before the breakdown, and the target vehicle's vehicle operating conditions after the breakdown repair. This can more concretely or explicitly express the relationship between historical maintenance data and vehicle breakdowns. Subsequently, based on the target vehicle's total historical breakdown count, the number of target vehicle breakdowns caused by battery cell failures, the target vehicle's parts replacement statistics, the fault codes of the faults that caused the target vehicle's breakdowns, the target vehicle's breakdown intervals, the target vehicle's vehicle operating data before the breakdown, and the target vehicle's vehicle operating conditions after the breakdown repair, it can more accurately predict whether the target vehicle has a breakdown risk. This allows the prediction model to identify the breakdown risk before the target vehicle breaks down (e.g., several hours or days before the breakdown occurs) and issue a breakdown risk warning, allowing users to intervene in a timely manner to ensure driving safety.

[0116] It should be noted that the statistical results of parts replacement of the target vehicle may include whether key components such as batteries, battery cells, and BMS have been replaced.

[0117] Of course, in other implementations, all anchoring-related data may be respectively taken as first anchoring-related data, and statistics may be performed on the first anchoring-related data to obtain data features of the first anchoring-related data.

[0118] In one embodiment, data features corresponding to multiple types of breakdown-related data are obtained separately. Specifically, at least one type of breakdown-related data is used as the second breakdown-related data, and the second breakdown-related data is directly used as the data feature of the second breakdown-related data. Because the second breakdown-related data itself can already embody or explicitly express its relationship with the vehicle breakdown, it is directly used as the corresponding data feature.

[0119] In one specific embodiment, the multiple types of breakdown-related data include vehicle operation data; if the vehicle operation data includes full vehicle operation data, the full vehicle operation data serves as the second breakdown-related data. The full vehicle operation data of the target vehicle can tangibly or explicitly express the relationship between the full vehicle operation data and vehicle breakdown. Subsequently, based on the full vehicle operation data, it is possible to accurately predict whether the target vehicle is at risk of breakdown. This allows the breakdown risk to be identified and issued before the target vehicle breaks down (e.g., hours or days before a breakdown occurs), enabling users to intervene in a timely manner to ensure driving safety.

[0120] In one specific embodiment, the multiple types of breakdown-related data include external environmental data. If the external environmental data includes geographic data, the geographic data serves as the second breakdown-related data. The geographic data of the target vehicle can tangibly or explicitly express the relationship between the target vehicle's geographic data and vehicle breakdown. Subsequently, based on the target vehicle's geographic data, whether the target vehicle is at risk of breakdown can be accurately predicted. This allows the breakdown risk to be identified and issued before the target vehicle breaks down (e.g., hours or days before a breakdown occurs), enabling timely user intervention to ensure driving safety.

[0121] Of course, in other implementations, all anchoring-related data may be used as second anchoring-related data, and the second anchoring-related data may be directly used as the data feature of the second anchoring-related data.

[0122] In one embodiment, multiple anchor-related data include time series data and non-time series data; before respectively obtaining the data features corresponding to the multiple anchor-related data, the time series data will be subjected to a first preprocessing; wherein, the first preprocessing includes at least one of the following: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data alignment; and, the non-time series data will be subjected to a second preprocessing; wherein, the second preprocessing includes at least one of the following: data format unification processing, missing data completion, and data alignment.

[0123] That is to say, a first preprocessing is performed on the time series data among the multiple anchoring-related data, and a second preprocessing is performed on the non-time series data among the multiple anchoring-related data, so as to convert the time series data and non-time series data among the multiple anchoring-related data into high-quality data. Subsequently, anchoring prediction is performed based on the preprocessed multiple anchoring-related data, which can improve the accuracy of anchoring prediction.

[0124] In a specific embodiment, the breakdown-related data includes vehicle operation data of the target vehicle, and the vehicle operation data is time series data; the time series data is subjected to a first preprocessing, specifically: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data time alignment.

[0125] In a specific embodiment, a signal processing algorithm (eg, moving average, Kalman filter, etc.) may be used to perform denoising on the vehicle operation data.

[0126] In one embodiment, for a small amount of randomly missing data, interpolation or average value can be used to fill in the missing data. For data that is continuously missing or has large gaps, a regression model can be used to predict and fill in the missing values to fill in the missing data.

[0127] In one embodiment, statistical methods (e.g., Z-score, IQR, etc.) can be used to detect outliers in vehicle operation data and process the outliers. Processing the outliers can include deleting the outliers, adjusting the outliers to the mean, median, etc., or marking the outliers so that they are not subsequently used.

[0128] In one embodiment, a logical consistency check is performed on the operating data of different vehicles within the same time period, and the logically inconsistent data is processed. The processing of the logically inconsistent data may include deleting the logically inconsistent data or adjusting the logically inconsistent data to logically consistent data.

[0129] For example, in a battery system, the battery pack of the target vehicle is composed of multiple battery cells, and the total voltage of the battery pack should theoretically be equal to the sum of the voltages of each battery cell.

[0130] For example, according to the principles of battery charging and discharging, when a battery is charging, current flows into the battery, and the battery's SOC increases with charging time; when the battery is discharging, current flows out of the battery, and the battery's SOC gradually decreases during the discharge process. Moreover, during the charging and discharging process, the magnitude of the current is correlated with the rate of change of the SOC. For example, in the initial charging phase, the greater the current, the faster the SOC rises; as the battery gradually reaches full capacity, the current gradually decreases, and the rate of SOC increase also slows. The opposite is true during the discharge process: the greater the discharge current, the faster the SOC decreases. Therefore, the change in current should theoretically match the trend of the SOC change.

[0131] In a specific embodiment, the vehicle operation data with different time stamps may be time-aligned using methods such as interpolation and resampling to ensure that different vehicle operation data are comparable at the same time point.

[0132] In a specific embodiment, the breakdown-related data includes historical maintenance data of the target vehicle, which is non-time series data; a second preprocessing is performed on the non-time series data, specifically: the historical maintenance data of the target vehicle is processed in a unified data format and missing data is completed.

[0133] Since the source channels of different historical maintenance data of the target vehicles are different (for example, the source channels may be after-sales service systems, maintenance reports, fault diagnosis systems, etc.), the recording formats of different historical maintenance data may be different. Therefore, it is necessary to unify the data formats of historical maintenance data.

[0134] In one specific implementation, techniques such as time series prediction and maintenance record correlation analysis can be used to supplement missing historical maintenance data. Furthermore, for missing important historical maintenance data, such as the cause of a vehicle breakdown or the repair measures, historical data from similar failures can be combined to infer the missing data.

[0135] In one specific embodiment, the breakdown-related data includes vehicle operation data of the target vehicle and data about the external environment in which the target vehicle is located. The external environment data includes weather data and geographic data. The vehicle operation data and weather data are time-series data, while the geographic data is non-time-series data. The step of preprocessing the breakdown-related data specifically comprises: aligning the vehicle operation data with the weather data in time; and aligning the vehicle operation data with the geographic data in space. In other words, the weather data is aligned with the vehicle operation data in time, and the geographic data is aligned with the vehicle operation data in space, to ensure that the external environment data of the target vehicle and the vehicle operation data of the target vehicle are aligned in the same time and space.

[0136] Since the characteristic scales of the data features corresponding to the various anchoring-related data may be different, in one embodiment, after respectively obtaining the data features corresponding to the various anchoring-related data, the data features corresponding to the various anchoring-related data are standardized or normalized.

[0137] Step S13: performing anchoring prediction based on data features corresponding to a plurality of anchoring-related data to obtain an anchoring prediction result.

[0138] In this embodiment, a breakdown prediction is performed based on data features corresponding to multiple types of breakdown-related data to obtain a breakdown prediction result. The breakdown prediction result includes a first prediction result, which is used to characterize whether the target vehicle is at risk of breakdown. The data features corresponding to the breakdown-related data can more concretely or explicitly express the relationship between the breakdown-related data and vehicle breakdown. Based on the data features corresponding to the breakdown-related data, the target vehicle's breakdown risk can be accurately predicted. Furthermore, the breakdown prediction based on the data features corresponding to multiple types of breakdown-related data can more comprehensively capture key information that may lead to the target vehicle's breakdown risk, thereby more accurately predicting the target vehicle's breakdown risk. Furthermore, the breakdown risk can be identified and a breakdown risk warning issued before the target vehicle breaks down (e.g., hours or days before the breakdown occurs), enabling timely user intervention to ensure driving safety.

[0139] In one embodiment, the first prediction result is used to indicate that the target vehicle has a breakdown risk. The breakdown prediction result also includes a second prediction result, which includes at least one of the following: a predicted time of occurrence of the breakdown risk, a predicted cause of the breakdown risk, and a solution to the breakdown risk. In other words, when a breakdown prediction is performed based on data features corresponding to multiple types of breakdown-related data and a breakdown risk is determined for the target vehicle, relevant information such as the time of occurrence of the breakdown risk, the cause of the breakdown risk, and the solution to the breakdown risk is also predicted. This provides a diverse and rich set of breakdown prediction results.

[0140] In one embodiment, after a breakdown prediction is performed based on data features corresponding to multiple breakdown-related data and a breakdown prediction result is obtained, at least a portion of the breakdown prediction result is fed back to the user. The feedback method includes at least one of the following: displaying at least a portion of the information on a display screen of the target vehicle, playing at least a portion of the information via audio on the target vehicle, or transmitting at least a portion of the information to an associated terminal. There are various methods for feeding back at least a portion of the breakdown prediction result to the user, and the feedback method can be flexibly selected.

[0141] When the feedback method is to display at least part of the information on the display screen of the target vehicle, at least part of the information in the breakdown prediction result can be presented in a visual manner, so as to display at least part of the information in the breakdown prediction result to the user in a concise and intuitive manner, so that the user can quickly understand the potential risks of the target vehicle, so that the user can take corresponding measures in time to ensure driving safety.

[0142] The information in at least part of the breakdown prediction result provided to the user is not limited. For example, a first prediction result may indicate that the target vehicle is at risk of breakdown. The first prediction result and a second prediction result may be provided to the user. The second prediction result may include the predicted time of occurrence of the breakdown risk, the predicted cause of the breakdown risk, and a solution to the breakdown risk.

[0143] In one embodiment, before performing anchoring prediction based on data features corresponding to multiple anchoring-related data and obtaining the anchoring prediction result, an associated feature group whose feature correlation meets a preset correlation condition is found from the data features corresponding to the multiple anchoring-related data, and a representative feature is selected from the associated feature group, and other data features in the associated feature group except the representative feature are deleted.

[0144] If the feature correlations between the data features in the associated feature group meet the preset correlation conditions, this indicates that the data features in the associated feature group are highly correlated and may contain information overlap. Therefore, representative features are selected from the associated feature group and retained, while other data features in the associated feature group are deleted. This effectively reduces redundant features in the data features corresponding to the various anchor correlation data, while retaining key features among the data features corresponding to the various anchor correlation data. In other words, redundant features are identified and deleted by quantifying the feature correlations between the data features corresponding to the various anchor correlation data.

[0145] The preset correlation condition is not limited. For example, the preset correlation condition is greater than or equal to a correlation threshold.

[0146] In a specific implementation, clustering may be used to find out the associated feature group whose feature correlation meets a preset correlation condition.

[0147] In other specific implementations, the feature correlation between any two data features may be calculated, and then an associated feature group whose feature correlation satisfies a preset correlation condition may be determined.

[0148] The MIC (Maximal Information Coefficient) value between each pair of data features can be calculated as the feature correlation between the two data features.

[0149] In one embodiment, before performing a breakdown prediction based on the data features corresponding to the various types of breakdown-related data and obtaining a breakdown prediction result, various time series feature sequences are retrieved from the data features corresponding to the various types of breakdown-related data, and each of these time series feature sequences is combined into a time series data feature. In other words, the time series feature sequences within the data features corresponding to the various types of breakdown-related data are combined to obtain a time series data feature to construct a combined feature. Combining the various time series feature sequences can enhance the ability to capture hidden relationships in the time series data, thereby improving the accuracy of subsequent vehicle breakdown predictions based on the time series data features.

[0150] In a specific embodiment, the time series feature sequence includes time series data corresponding to different moments. The time series data corresponding to the same moment can also be combined as the time series data feature corresponding to the moment. In this case, multiple combined features will be constructed (one time series data feature corresponds to one moment).

[0151] In one embodiment, before performing a breakdown prediction based on the data features corresponding to the various types of breakdown-related data and obtaining a breakdown prediction result, non-time-series data is searched for from the data features corresponding to the various types of breakdown-related data, and the non-time-series data is combined to obtain non-time-series data features. Combining non-time-series data can enhance the ability to capture hidden relationships in the non-time-series data, thereby improving the accuracy of subsequent vehicle breakdown predictions based on the non-time-series data features.

[0152] In one embodiment, the vehicle breakdown warning method provided by this application can be executed locally. Of course, in other embodiments, the vehicle breakdown warning method provided by this application can also be executed in the cloud. The cloud service can also provide nationwide vehicle breakdown risk prediction analysis to help enterprises or service providers understand the health status of vehicles in advance.

[0153] See also Figure 2 , Figure 2 yes Figure 1 It should be noted that if there is substantially the same result, this embodiment does not use Figure 2 The process sequence shown is limited. Figure 2 As shown, the data features corresponding to the various anchor-related data include time series data features and non-time series data features. This embodiment includes:

[0154] Step S21: Utilize the time series processing model to detect the time series data feature dependencies in time series to obtain time-dependent features.

[0155] In this implementation, a time series processing model is used to detect the temporal dependencies of time series data features to obtain time-dependent features. Specifically, the time series processing model is used to model the input time series data features and extract high-order representations that reflect the dynamic changes in the time dimension, namely time-dependent features. This captures the long-term temporal dependencies of time series data features, thereby providing key features for subsequent vehicle breakdown prediction.

[0156] In one embodiment, the time series processing model is a long short-term memory network model (LSTM). Of course, in other embodiments, the time series processing model can also be other models, which is not limited here.

[0157] Step S22: using the decision model to identify the association between the non-time series data features and the anchor to obtain the anchor association features.

[0158] In this embodiment, a decision model is used to identify the association between non-time series data features and breakdowns to obtain breakdown-related features. In other words, the decision model can identify breakdown-related features associated with breakdown risk from non-time series data features. Specifically, the decision model can accurately locate breakdown risk factors from non-time series data features, thereby providing key feature support for subsequent vehicle breakdown prediction.

[0159] In one embodiment, a decision model may be used to identify a single anchoring-related feature that is most associated with anchoring risk from time series data features. Of course, in other embodiments, a decision model may be used to identify multiple anchoring-related features associated with anchoring risk from time series data features, which is not limited here.

[0160] In one embodiment, the decision model is a random forest model (RF). Of course, in other embodiments, the decision model may also be a LASSO model, which is not limited here.

[0161] Step S23: Use the time-dependent features and the anchoring-related features to make predictions and obtain anchoring prediction results.

[0162] In this implementation, time-dependent features and breakdown-related features are used to predict vehicle breakdowns. Time-dependent features are derived by using a time series processing model to detect the temporal dependencies of time series data features. They can reflect the long-term decay trends (e.g., battery capacity decline) and short-term abnormal fluctuations (e.g., battery temperature surges) of the target vehicle's time series data, modeling its dynamic evolution. Breakdown-related features are derived by using a decision model to identify the association between non-time series data features and breakdowns, enabling exploration of discrete event correlations. Therefore, using both time-dependent and breakdown-related features for prediction simultaneously covers both "gradual aging" and "sudden impact" risks, avoiding the blind spots of a single data source and improving the accuracy of vehicle breakdown risk prediction.

[0163] In one embodiment, a fully connected layer is used to perform prediction based on time-dependent features and anchor-related features to obtain an anchor prediction result.

[0164] In one embodiment, the anchoring prediction result is obtained by using the time-dependent features and the anchoring correlation features for prediction. Specifically, the time-dependent features and the anchoring correlation features are fused to obtain a fused feature; and prediction is performed based on the fused feature to obtain the anchoring prediction result. In other words, the time-dependent features and the anchoring correlation features are fused, and anchoring prediction is performed based on the fused feature to obtain the anchoring prediction result. By fusing the time-dependent features with the anchoring correlation features, a multi-dimensional and multi-granular anchoring risk characterization system is constructed, thereby improving the accuracy of anchoring risk prediction.

[0165] In a specific implementation, the fusion of the time-dependent feature and the anchoring correlation feature may be to concatenate the time-dependent feature and the anchoring correlation feature.

[0166] In one specific embodiment, the time series processing model is a long short-term memory network model; the decision model is a random forest model; a fully connected layer is used to make predictions based on time-dependent features and anchoring correlation features to obtain anchoring prediction results. The long short-term memory network model and the random forest model are trained separately, wherein a binary cross-entropy loss function is used to adjust the model's hyperparameters through grid search, such as the number of layers and hidden units of the long short-term memory network model, the number of trees of the random forest model, etc., to improve model performance; then, the performance of the long short-term memory network model and the random forest model is evaluated using accuracy, recall rate, F1 score, etc., and the model is verified using cross-validation to prevent overfitting. After completing the training of the long short-term memory network model and the random forest model, the fully connected layer is trained.

[0167] In other embodiments, a multi-input-output model can also be used to perform anchoring prediction based on data features corresponding to a variety of anchoring-related data to obtain an anchoring prediction result. Specifically, the multi-input-output model includes a time series branch, a non-time series branch, and a prediction branch. The time series branch, the non-time series branch, and the prediction branch work together to form a multi-input-output model that can perform anchoring prediction based on data features corresponding to a variety of anchoring-related data. The time series branch can be regarded as a time series processing layer for processing time series data in the multi-input-output model. The time series branch detects the time series dependency of time series data features to obtain time-dependent features; the non-time series branch can be regarded as a non-time series processing layer for processing non-time series data in the multi-input-output model. The non-time series branch identifies the association between non-time series data features and anchoring to obtain anchoring-related features; the prediction branch can be regarded as a prediction layer for anchoring prediction in the multi-input-output model. The prediction branch uses time-dependent features and anchoring-related features to perform prediction and obtain an anchoring prediction result.

[0168] Among them, the time series processing layer can be an LSTM layer, the non-time series processing layer can be an RF layer, and the prediction layer can be a fully connected layer.

[0169] See also Figure 3 , Figure 3 : It is a structural diagram of an embodiment of a vehicle breakdown warning device provided by the present application. The vehicle breakdown warning device 30 includes a first acquisition module 31, a second acquisition module 32 and a prediction module 33. The first acquisition module 31 is used to obtain a variety of breakdown-related data of the target vehicle, wherein the multiple breakdown-related data include historical maintenance data of the target vehicle, at least one of the external environment data of the target vehicle, and vehicle operation data of the target vehicle; the second acquisition module 32 is used to respectively obtain data features corresponding to the multiple breakdown-related data; the prediction module 33 is used to perform breakdown prediction based on the data features corresponding to the multiple breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to characterize whether the target vehicle has a breakdown risk.

[0170] Among them, the above-mentioned vehicle operation data includes at least one of the following: battery cell voltage data, battery cell current data, vehicle temperature data, battery pack charge status data, and vehicle operation data; historical maintenance data includes at least one of the following: maintenance time, vehicle breakdown reason, maintenance content, vehicle status data before and after the failure, and the external environment conditions of the target vehicle when it broke down; external environment data includes at least one of weather data and geographic data; among them, vehicle operation data and weather data are time series data.

[0171] Among them, the above-mentioned battery cell voltage data includes the voltage value of each battery cell in the battery pack of the target vehicle at different times; the battery cell current data includes the charging and discharging current of each battery cell at different times; the vehicle temperature data includes at least one of the following: the battery cell temperature at different times, the internal ambient temperature of the target vehicle at different times; the whole vehicle operation data includes at least one of the following: the vehicle speed of the target vehicle at different times, the power output of the target vehicle at different times; the vehicle status data before and after the fault includes at least one of the following: the vehicle operation status before and after the fault, the vehicle sensor data before and after the fault; the weather data includes at least one of the following: the external ambient temperature corresponding to different times, the external ambient humidity corresponding to different times; the geographical data includes road condition information.

[0172] Among them, the second acquisition module 32 is used to respectively obtain data features corresponding to multiple anchoring-related data, including: taking at least one anchoring-related data as the first anchoring-related data, and performing statistics on the first anchoring-related data to obtain data features of the first anchoring-related data; and / or, the second acquisition module 32 is used to respectively take at least one anchoring-related data as the second anchoring-related data, and directly take the second anchoring-related data as the data features of the second anchoring-related data.

[0173] Among them, the above-mentioned multiple breakdown-related data include vehicle operation data; in the case that the vehicle operation data includes battery cell voltage data, the battery cell voltage data is the first breakdown-related data, and the data characteristics of the battery cell voltage data include at least one of the following: the first battery cell voltage difference, the second battery cell voltage difference, the voltage fluctuation characterization value corresponding to each battery cell, and the number of voltage anomalies of the battery pack within the first time window, wherein the first battery cell voltage difference is the voltage difference between different battery cells, and the second battery cell voltage difference is the difference between the sum of the voltages of each battery cell and the total voltage of the battery pack; in the case that the vehicle operation data includes battery cell current data, the battery cell current data is the first breakdown-related data, and the data characteristics of the battery cell current data include at least one of the following: the discharge current fluctuation characterization value corresponding to each battery cell, the number of voltage anomalies of the battery pack within the first time window The number of charging current anomalies within the second time window, and the correlation between the current and voltage of each battery cell; when the vehicle operation data includes vehicle temperature data, the vehicle temperature data is the first breakdown-related data, and the data characteristics of the vehicle temperature data include at least one of the following: the temperature distribution difference of each battery cell, the temperature fluctuation characterization value of each battery cell, and the number of temperature anomalies of the battery pack within the third time window; when the vehicle operation data includes the state of charge data of the battery pack, the state of charge data is the first breakdown-related data, and the data characteristics of the state of charge data include at least one of the following: the charge fluctuation characterization value of the battery pack, the relationship between the charge consumption value of the battery pack and the mileage of the target vehicle; when the vehicle operation data includes the whole vehicle operation data, the whole vehicle operation data is the second breakdown-related data.

[0174] Among them, the above-mentioned vehicle operation data is time series data, and the data characteristics corresponding to each vehicle operation data are time series characteristic sequences, among which the time series characteristic sequences corresponding to the cell voltage data include the first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell and the number of voltage anomalies at different times; the time series characteristic sequences corresponding to the cell current data include the discharge current fluctuation characterization value corresponding to different times, the number of charging current anomalies, and the correlation between the current and voltage of each cell; the time series characteristic sequences corresponding to the vehicle temperature data include the temperature distribution difference, temperature fluctuation characterization value and the number of temperature anomalies corresponding to different times; the time series characteristic sequences corresponding to the state of charge data include the temperature distribution difference, temperature fluctuation characterization value and the number of temperature anomalies corresponding to different times. The timing feature sequence includes the charge fluctuation characterization values corresponding to different moments, the relationship between the charge value of the battery pack and the mileage of the target vehicle; and / or, the voltage fluctuation characterization value corresponding to the battery cell includes the voltage change rate of the battery cell; and / or, the first battery cell voltage difference includes the voltage difference between the maximum voltage battery cell and the minimum voltage battery cell; and / or, the discharge current fluctuation characterization value corresponding to the battery cell includes the discharge current change rate of the battery cell; and / or, the temperature distribution difference of the battery cell includes the temperature difference between the maximum temperature battery cell and the minimum temperature battery cell; and / or, the temperature fluctuation characterization value of the battery cell includes the temperature change rate of the battery cell; and / or, the charge fluctuation characterization value of the battery pack includes the charge decrease rate of the battery pack.

[0175] Among them, when the anchoring-related data includes external environment data, and the external environment data includes the external environment temperature, the above-mentioned external environment temperature is the first anchoring-related data, and the data characteristics of the external environment temperature include at least one of the following: the external environment temperature central trend statistical value, the external environment temperature fluctuation characterization value, and the external environment temperature distribution within the fourth time window; when the anchoring-related data includes external environment data, and the external environment data includes the external environment humidity, the external environment humidity is the first anchoring-related data, and the data characteristics of the external environment humidity include at least one of the following: the external environment humidity central trend statistical value, the external environment humidity fluctuation characterization value; when the anchoring-related data includes the historical maintenance data of the target vehicle, the historical maintenance data is the first anchoring-related data, and the data characteristics of the historical maintenance data include at least one of the following: the total number of historical anchoring times of the target vehicle, the number of times the target vehicle has been anchored due to cell failure, the statistical results of parts replacement of the target vehicle, the fault code of the fault that caused the target vehicle to be anchored, the anchoring interval of the target vehicle, the vehicle operation data of the target vehicle before the anchoring, and the vehicle operation status of the target vehicle after the anchoring repair.

[0176] Among them, the above-mentioned external ambient temperature and external ambient humidity are time series data, and the data features corresponding to the external ambient temperature and the external ambient humidity are time series feature sequences, wherein the time series feature sequence corresponding to the external ambient temperature includes the external ambient temperature central trend statistical values corresponding to different moments, the external ambient temperature fluctuation characterization values, and the external ambient temperature distribution; the time series feature sequence corresponding to the external ambient humidity includes the external ambient humidity central trend statistical values and the external ambient humidity fluctuation characterization values corresponding to different moments; and / or, the external ambient temperature fluctuation characterization values include the external ambient temperature change rate; and / or, the external ambient temperature distribution includes the highest temperature and the lowest temperature within the fourth time window; and / or, the external ambient humidity fluctuation characterization values include the external ambient humidity change rate.

[0177] Among them, the prediction module 33 is also used to perform anchoring prediction based on the data features corresponding to the multiple anchoring-related data. Before obtaining the anchoring prediction result, it is used to find out the associated feature group whose feature correlation meets the preset correlation condition from the data features corresponding to the multiple anchoring-related data, select the representative feature from the associated feature group, and delete other data features in the associated feature group except the representative feature; find out each time series feature sequence from the data features corresponding to the multiple anchoring-related data, and combine each time series feature sequence into a time series data feature.

[0178] Among them, the data features corresponding to the above-mentioned multiple anchoring-related data respectively include time series data features and non-time series data features; the prediction module 33 is used to perform anchoring prediction based on the data features corresponding to the multiple anchoring-related data to obtain anchoring prediction results, including: using a time series processing model to detect the time series dependency of time series data features to obtain time-dependent features; and using a decision model to identify the association between non-time series data features and anchoring to obtain anchoring-related features; using time-dependent features and anchoring-related features to perform prediction to obtain anchoring prediction results.

[0179] Among them, the above-mentioned time series processing model is a long short-term memory network model, and the decision model is a random forest model; and / or, the prediction module 33 is used to use time-dependent features and anchoring-related features to make predictions to obtain the anchoring prediction results, including: fusing time-dependent features and anchoring-related features to obtain fused features; making predictions based on fused features to obtain anchoring prediction results.

[0180] Among them, the above-mentioned multiple anchor-related data include time series data and non-time series data; the second acquisition module 32 is also used to perform a first preprocessing on the time series data before respectively obtaining the data features corresponding to the multiple anchor-related data; wherein, the first preprocessing includes at least one of the following: denoising, missing data completion, abnormal data processing, inconsistent data processing, data alignment; and, performing a second preprocessing on the non-time series data; wherein, the second preprocessing includes at least one of the following: data format unification processing, missing data completion, and data alignment.

[0181] Among them, the above-mentioned breakdown-related data includes vehicle operation data of the target vehicle, and the vehicle operation data is time series data; the second acquisition module 32 is used to perform a first preprocessing on the time series data, including: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data time alignment of the vehicle operation data respectively; and / or, the above-mentioned breakdown-related data includes historical maintenance data of the target vehicle, and the historical maintenance data is non-time series data; the second acquisition module 32 is used to perform a second preprocessing on the non-time series data, including: data format unification processing and missing data completion of the historical maintenance data of the target vehicle respectively; and / or, the above-mentioned breakdown-related data includes vehicle operation data of the target vehicle and external environment data of the target vehicle, the external environment data includes weather data and geographic data, the vehicle operation data and weather data are time series data, and the geographic data is non-time series data; the second acquisition module 32 is used to perform a step of preprocessing the breakdown-related data, and also includes: data time alignment of the vehicle operation data and the weather data; and data space alignment of the vehicle operation data and the geographic data.

[0182] Among them, the above-mentioned first prediction result is used to characterize the risk of breakdown of the target vehicle, and the breakdown prediction result also includes a second prediction result, and the second prediction result includes at least one of the following: the predicted occurrence time of the breakdown risk, the predicted cause of the breakdown risk, and the solution to the breakdown risk; and / or, the vehicle breakdown warning device 30 also includes a feedback module 34, and the feedback module 34 is used to perform an anchoring prediction based on data features corresponding to a plurality of anchoring-related data, and after obtaining the anchoring prediction result, feedback at least part of the information in the anchoring prediction result to the user, wherein the feedback method includes at least one of the following: displaying at least part of the information on the display screen of the target vehicle, playing at least part of the information on the audio on the target vehicle, and sending at least part of the information to the associated terminal.

[0183] See also Figure 4 , Figure 4is a schematic diagram of the structure of an embodiment of an electronic device provided herein. Electronic device 40 includes a memory 41 and a processor 42 coupled to each other. Processor 42 is configured to execute program instructions stored in memory 41 to implement the steps of any of the aforementioned vehicle breakdown warning method embodiments. In a specific implementation scenario, electronic device 40 may include, but is not limited to, a microcomputer and a server. Furthermore, electronic device 40 may also include mobile devices such as laptops and tablet computers, without limitation herein.

[0184] Specifically, the processor 42 is used to control itself and the memory 41 to implement the steps of any of the above-mentioned vehicle breakdown warning method embodiments. The processor 42 can also be referred to as a CPU (Central Processing Unit). The processor 42 may be an integrated circuit chip with signal processing capabilities. The processor 42 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor. In addition, the processor 42 can be implemented by an integrated circuit chip.

[0185] See also Figure 5 , Figure 5 The figure is a schematic diagram of the structure of an embodiment of a computer-readable storage medium provided by the present application. The computer-readable storage medium 50 of this embodiment of the present application stores program instructions 51. When executed, these program instructions 51 implement the method provided by any embodiment of the vehicle breakdown warning method of the present application, as well as any non-conflicting combination thereof. The program instructions 51 can be stored in the computer-readable storage medium 50 as a program file in the form of a software product, enabling a computer device (such as a personal computer, server, or network device) to execute all or part of the steps of the method of each embodiment of the present application. The computer-readable storage medium 50 includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or terminal devices such as a computer, server, mobile phone, or tablet.

[0186] The above description is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A vehicle breakdown warning method, characterized in that: The method comprises: Acquiring a plurality of breakdown-related data of a target vehicle, wherein the plurality of breakdown-related data includes at least one of historical maintenance data of the target vehicle, external environment data of the target vehicle, and vehicle operation data of the target vehicle; respectively obtaining data features corresponding to the plurality of anchoring-related data; Performing a breakdown prediction based on data features corresponding to the multiple breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to indicate whether the target vehicle has a breakdown risk; The data features corresponding to the multiple anchoring-related data include time series data features and non-time series data features; and performing anchoring prediction based on the data features corresponding to the multiple anchoring-related data to obtain an anchoring prediction result includes: Using a time series processing model to detect the time series data feature's time series dependency to obtain a time-dependent feature; and Using a decision model to identify the association between the non-time series data feature and the anchor, so as to obtain an anchor association feature; Fusing the time-dependent feature and the anchoring-related feature to obtain a fused feature; Prediction is performed based on the fusion features to obtain the anchor prediction result.

2. The method according to claim 1, characterized in that The vehicle operation data includes at least one of the following: battery cell voltage data, battery cell current data, vehicle temperature data, battery pack state of charge data, and vehicle operation data; The historical maintenance data includes at least one of the following: maintenance time, vehicle breakdown reason, maintenance content, vehicle status data before and after the breakdown, and the external environment of the target vehicle when the breakdown occurred; The external environment data includes at least one of weather data and geographical data; Among them, the vehicle operation data and weather data are time series data.

3. The method according to claim 2, characterized in that The cell voltage data includes the voltage value of each cell in the battery pack of the target vehicle at different times; The cell current data includes the charge and discharge current of each cell at different times; The vehicle temperature data includes at least one of the following: battery core temperature at different times, and internal ambient temperature of the target vehicle at different times; The vehicle operation data includes at least one of the following: the vehicle speed of the target vehicle at different times, the power output of the target vehicle at different times; The vehicle status data before and after the fault includes at least one of the following: vehicle operating status before and after the fault, vehicle sensor data before and after the fault; The weather data includes at least one of the following: external environment temperatures corresponding to different times, and external environment humidity corresponding to different times; The geographic data includes road condition information.

4. The method according to any one of claims 1 to 3, characterized in that The respectively obtaining data features corresponding to the multiple types of anchoring-related data includes: taking at least one of the anchoring-related data as first anchoring-related data, and performing statistics on the first anchoring-related data to obtain data features of the first anchoring-related data; and / or, At least one of the anchoring-related data is used as the second anchoring-related data, and the second anchoring-related data is directly used as the data feature of the second anchoring-related data.

5. The method according to claim 4, characterized in that The plurality of breakdown-related data includes the vehicle operation data; In a case where the vehicle operation data includes cell voltage data, the cell voltage data is the first breakdown-related data, and data features of the cell voltage data include at least one of the following: a first cell voltage difference, a second cell voltage difference, a voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies of the battery pack within a first time window, wherein the first cell voltage difference is a voltage difference between different cells, and the second cell voltage difference is a difference between the sum of the voltages of the cells and the total voltage of the battery pack; In a case where the vehicle operation data includes cell current data, the cell current data is the first breakdown-related data, and data characteristics of the cell current data include at least one of the following: a discharge current fluctuation characteristic value corresponding to each cell, the number of abnormal charging currents of the battery pack within the second time window, and a correlation between the current and voltage of each cell; In a case where the vehicle operation data includes vehicle temperature data, the vehicle temperature data is the first breakdown-related data, and data characteristics of the vehicle temperature data include at least one of the following: a temperature distribution difference of each battery cell, a temperature fluctuation characteristic value of each battery cell, and a number of temperature anomalies of the battery pack within a third time window; In a case where the vehicle operation data includes state of charge data of a battery pack, the state of charge data is the first breakdown-related data, and the data characteristics of the state of charge data include at least one of the following: a charge fluctuation characterization value of the battery pack, a relationship between a charge consumption value of the battery pack and the mileage of the target vehicle; In a case where the vehicle operation data includes whole vehicle operation data, the whole vehicle operation data is the second breakdown-related data.

6. The method according to claim 5, characterized in that The vehicle operation data is time series data, and the data features corresponding to each of the vehicle operation data are time series feature sequences, wherein the time series feature sequences corresponding to the cell voltage data include the first cell voltage difference, the second cell voltage difference, the voltage fluctuation characterization value corresponding to each cell, and the number of voltage anomalies corresponding to different moments; the time series feature sequences corresponding to the cell current data include the discharge current fluctuation characterization value, the number of charging current anomalies, and the correlation between the current and voltage of each cell corresponding to different moments; the time series feature sequences corresponding to the vehicle temperature data include the temperature distribution difference, the temperature fluctuation characterization value, and the number of temperature anomalies corresponding to different moments; the time series feature sequences corresponding to the state of charge data include the charge fluctuation characterization value corresponding to different moments, the relationship between the charge value of the battery pack and the mileage of the target vehicle; And / or, the voltage fluctuation characterization value corresponding to the battery cell includes the voltage change rate of the battery cell; And / or, the first cell voltage difference includes a voltage difference between a cell with a maximum voltage and a cell with a minimum voltage; And / or, the discharge current fluctuation characterization value corresponding to the battery cell includes the discharge current change rate of the battery cell; And / or, the temperature distribution difference of the battery cells includes a temperature difference between a battery cell with a maximum temperature and a battery cell with a minimum temperature; And / or, the temperature fluctuation characterization value of the battery cell includes the temperature change rate of the battery cell; And / or, the charge fluctuation characterization value of the battery pack includes a charge decrease rate of the battery pack.

7. The method according to claim 4, characterized in that In a case where the anchoring-related data includes external environment data, and the external environment data includes external environment temperature, the external environment temperature is the first anchoring-related data, and the data characteristics of the external environment temperature include at least one of the following: a central tendency statistical value of the external environment temperature, a characteristic value of external environment temperature fluctuation, and a distribution of the external environment temperature within a fourth time window; In a case where the anchoring-related data includes external environment data, and the external environment data includes external environment humidity, the external environment humidity is the first anchoring-related data, and the data characteristics of the external environment humidity include at least one of the following: a central tendency statistical value of the external environment humidity, a fluctuation characterization value of the external environment humidity; In the case where the breakdown-related data includes historical maintenance data of the target vehicle, the historical maintenance data is the first breakdown-related data, and the data characteristics of the historical maintenance data include at least one of the following: the total number of historical breakdowns of the target vehicle, the number of times the target vehicle breaks down due to battery cell failure, the statistical results of parts replacement of the target vehicle, the fault code of the fault that caused the target vehicle to break down, the breakdown interval of the target vehicle, the vehicle operation data of the target vehicle before the breakdown, and the vehicle operation status of the target vehicle after the breakdown repair.

8. The method according to claim 7, characterized in that The external environment temperature and the external environment humidity are time series data, and the data features corresponding to the external environment temperature and the external environment humidity are time series feature sequences, wherein the time series feature sequences corresponding to the external environment temperature include the external environment temperature central tendency statistical values corresponding to different moments, the external environment temperature fluctuation characterization values, and the external environment temperature distribution; the time series feature sequences corresponding to the external environment humidity include the external environment humidity central tendency statistical values and the external environment humidity fluctuation characterization values corresponding to different moments; And / or, the external environment temperature fluctuation characterization value includes the external environment temperature change rate; And / or, the external environment temperature distribution includes the highest temperature and the lowest temperature within a fourth time window; And / or, the external environment humidity fluctuation characterization value includes the external environment humidity change rate.

9. The method according to any one of claims 1 to 3, characterized in that Before performing anchoring prediction based on the data features corresponding to the multiple anchoring-related data to obtain an anchoring prediction result, at least one of the following steps is further included: Finding, from the data features corresponding to the plurality of anchor-related data, a correlation feature group whose feature correlation satisfies a preset correlation condition, selecting a representative feature from the correlation feature group, and deleting other data features in the correlation feature group except the representative feature; Find out each time series feature sequence from the data features corresponding to the plurality of anchor-related data, and combine each of the time series feature sequences into a time series data feature.

10. The method according to claim 1, characterized in that The time series processing model is a long short-term memory network model, and the decision model is a random forest model.

11. The method according to claim 1, wherein The multiple types of anchor-related data include time series data and non-time series data; and before respectively obtaining data features corresponding to the multiple types of anchor-related data, the method further includes: Performing a first preprocessing on the time series data; wherein the first preprocessing includes at least one of the following: denoising, missing data completion, abnormal data processing, inconsistent data processing, and data alignment; And, performing a second preprocessing on the non-time series data; wherein, the second preprocessing includes at least one of the following: data format unification processing, missing data completion, and data alignment.

12. The method according to claim 11, characterized in that The breakdown-related data includes vehicle operation data of the target vehicle, and the vehicle operation data is the time series data; and performing a first preprocessing on the time series data includes: De-noising, missing data completion, abnormal data processing, inconsistent data processing, and data time alignment are performed on the vehicle operation data respectively; And / or, the breakdown-related data includes historical maintenance data of the target vehicle, and the historical maintenance data is the non-time series data; and the second preprocessing of the non-time series data includes: Performing data format unification and missing data completion on the historical maintenance data of the target vehicle; And / or, the breakdown-related data includes vehicle operation data of the target vehicle and external environment data of the target vehicle, the external environment data includes weather data and geographic data, the vehicle operation data and the weather data are the time series data, and the geographic data is the non-time series data; the step of preprocessing the breakdown-related data further includes: Performing data time alignment on the vehicle operation data and the weather data; and The vehicle operation data is spatially aligned with the geographic data.

13. The method according to claim 1, wherein The first prediction result is used to indicate that the target vehicle has a breakdown risk. The breakdown prediction result further includes a second prediction result, wherein the second prediction result includes at least one of the following: a predicted occurrence time of the breakdown risk, a predicted cause of the breakdown risk, and a solution to the breakdown risk; And / or, after performing anchoring prediction based on the data features corresponding to the multiple anchoring-related data to obtain an anchoring prediction result, the method further includes: Feedback at least part of the breakdown prediction result to the user, wherein the feedback method includes at least one of the following: displaying at least part of the information on a display screen of the target vehicle, playing at least part of the information through audio on the target vehicle, and sending at least part of the information to an associated terminal.

14. A vehicle breakdown warning device, characterized in that: The device comprises: a first acquisition module, configured to acquire a plurality of breakdown-related data of a target vehicle, wherein the plurality of breakdown-related data includes at least one of historical maintenance data of the target vehicle, external environment data of the target vehicle, and vehicle operation data of the target vehicle; A second acquisition module is used to respectively acquire data features corresponding to the multiple types of anchoring-related data; A prediction module, configured to perform a breakdown prediction based on data features corresponding to the plurality of breakdown-related data to obtain a breakdown prediction result; wherein the breakdown prediction result includes a first prediction result, and the first prediction result is used to indicate whether the target vehicle has a breakdown risk; The data features corresponding to the multiple anchoring-related data include time series data features and non-time series data features; and performing anchoring prediction based on the data features corresponding to the multiple anchoring-related data to obtain an anchoring prediction result includes: Using a time series processing model to detect the time series data feature's time series dependency to obtain a time-dependent feature; and Using a decision model to identify the association between the non-time series data feature and the anchor, so as to obtain an anchor association feature; Fusing the time-dependent feature and the anchoring-related feature to obtain a fused feature; Prediction is performed based on the fusion features to obtain the anchor prediction result.

15. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores program instructions, and the processor is used to execute the program instructions to implement the vehicle breakdown warning method according to any one of claims 1 to 13.

16. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program instructions, and the program instructions can be executed to implement the vehicle breakdown warning method according to any one of claims 1 to 13.

Images