Vehicle fault prediction method and device and storage medium

Abstract

The invention relates to a vehicle fault prediction method and device and a storage medium, and the method comprises the steps: obtaining multi-source heterogeneous data of a vehicle, the multi-source heterogeneous data comprising at least two types of equipment sensing data, driving data and environment data of the vehicle; performing feature extraction on different data in the multi-source heterogeneous data based on a mode corresponding to each type to obtain a plurality of feature vectors; and obtaining the fault risk of the vehicle in the future time period through a prediction model based on the plurality of feature vectors. According to the method, by integrating the equipment sensing data, the driving data and the environment data, information richer than that of a single data source can be provided, the running state of the vehicle can be comprehensively evaluated, and therefore potential faults can be predicted more accurately.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a vehicle fault prediction method, device, and storage medium. Background Art

[0002] As modern vehicle systems become increasingly complex, accurately predicting potential vehicle failures can improve driving safety. Related technologies rely on on-board diagnostics (OBD) systems to obtain limited fault code information, which often fails to fully reflect the vehicle's operating status, resulting in inaccurate predictions. Summary of the Invention

[0003] The embodiments of the present application provide a vehicle fault prediction method, device, and storage medium to solve the above-mentioned problems.

[0004] To achieve the above objectives, according to a first aspect of the present application, a vehicle fault prediction method is provided, the method comprising:

[0005] Acquiring multi-source heterogeneous data of a vehicle, wherein the multi-source heterogeneous data includes at least two types of data selected from the group consisting of device sensor data, driving data, and environmental data of the vehicle;

[0006] Extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each of the types to obtain multiple feature vectors;

[0007] Based on the multiple feature vectors, a prediction model is used to obtain the failure risk of the vehicle in a future time period.

[0008] Optionally, the multi-source heterogeneous data includes first driving data, second driving data and third driving data, and the multiple feature vectors include driving behavior features.

[0009] The method of extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes:

[0010] When the frequency of the first driving data being greater than or equal to the preset data threshold is greater than the preset frequency threshold, and / or

[0011] The duration of the second driving data being greater than or equal to the preset data threshold is greater than a preset time threshold, and / or

[0012] When the third driving data is greater than or equal to the preset data threshold,

[0013] The driving behavior characteristic is determined based on the first driving data, the second driving data and / or the third driving data.

[0014] Optionally, the first driving data includes inertial motion data of the vehicle, and the driving behavior characteristics corresponding to the first driving data include at least one of acceleration information, start-stop information, and angular velocity information, and / or

[0015] The second driving data includes positioning data of the vehicle, and the driving behavior characteristics corresponding to the second driving data include speeding information, and / or

[0016] The third driving data includes bus data of the vehicle, and the driving behavior characteristics corresponding to the third driving data include starting vibration information and / or braking intensity information.

[0017] Optionally, the device sensor data includes at least one of battery data, engine data, fault code number, and fuel tank data, and the multiple feature vectors include fault features.

[0018] The method of extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes:

[0019] When the number of fault codes is greater than or equal to a preset number threshold; and / or

[0020] When the battery data is abnormal; and / or the engine data is abnormal; and / or the fuel tank data is abnormal;

[0021] The fault characteristic is determined based on at least one of the fault code number, the battery data, the engine data, and the fuel tank data.

[0022] Optionally, the feature vector further includes environmental features, and the input of the prediction model further includes a driving behavior score, which is determined by:

[0023] A causal relationship graph is constructed based on the driving behavior characteristics, the fault characteristics, and the environmental characteristics; the causal relationship graph is graph structure data, and the graph structure data includes nodes and edges; each node corresponds to a type of the multi-source heterogeneous data, and the characteristics of the node include a feature vector corresponding to the multi-source heterogeneous data; the edges are related to the causal relationship of the fault occurrence, and the characteristics of the edges include the causal strength of the causal relationship;

[0024] Based on the cause-effect relationship diagram, the driving behavior score is determined by a preset scoring model.

[0025] Optionally, the extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes:

[0026] Preprocessing the multi-source heterogeneous data based on the collection characteristics of the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data;

[0027] Feature extraction is performed on the plurality of preprocessed multi-source heterogeneous data to obtain the plurality of feature vectors.

[0028] Optionally, the acquisition feature includes a sampling frequency, and preprocessing the multi-source heterogeneous data based on the acquisition feature of the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data includes:

[0029] Based on the first multi-source heterogeneous data with the highest sampling frequency, the second multi-source heterogeneous data is resampled to obtain the resampled second multi-source heterogeneous data that is aligned with the first multi-source heterogeneous data at multiple time points, so as to obtain the preprocessed multi-source heterogeneous data, wherein the second multi-source heterogeneous data is the multi-source heterogeneous data other than the first multi-source heterogeneous data.

[0030] Optionally, the performing feature extraction on the plurality of pre-processed multi-source heterogeneous data to obtain the plurality of feature vectors includes:

[0031] Sliding a sliding window based on a preset time length on the pre-processed multi-source heterogeneous data according to a preset step length to divide the pre-processed multi-source heterogeneous data into a plurality of interval data;

[0032] Feature extraction is performed on the statistical features of the multiple interval data to obtain feature vectors corresponding to the multi-source heterogeneous data.

[0033] According to a second aspect of the present application, an embodiment of the present application further provides a vehicle fault prediction device, the device comprising:

[0034] an acquisition module, configured to acquire multi-source heterogeneous data of a vehicle, wherein the multi-source heterogeneous data includes at least two types of data selected from the group consisting of device sensor data, driving data, and environmental data of the vehicle;

[0035] an extraction module, configured to extract features from different data in the multi-source heterogeneous data based on a method corresponding to each of the types, to obtain a plurality of feature vectors;

[0036] The prediction module is used to obtain the failure risk of the vehicle in a future time period based on the multiple feature vectors through a prediction model.

[0037] According to the third aspect of the present application, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the methods provided in the embodiments of the present application.

[0038] Some embodiments of this specification include at least the following beneficial effects: by integrating multi-source heterogeneous data of the vehicle, including equipment sensor data, driving data, and environmental data, and based on different feature extraction methods, multiple feature vectors are extracted and input into the prediction model, which can not only realize the identification of potential vehicle faults, but also realize the prediction of the probability of fault occurrence in the future time period, thereby improving the accuracy of fault prediction.

[0039] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0041] In order to more completely understand the present application and its beneficial effects, the following description will be given in conjunction with the accompanying drawings, wherein the same drawing numbers represent the same parts in the following description.

[0042] Figure 1 is an application scenario diagram of the vehicle fault prediction method according to some embodiments of this specification;

[0043] Figure 2 is an exemplary flow chart of a vehicle fault prediction method according to some embodiments of this specification;

[0044] Figure 3 is an exemplary flow chart of determining a driving behavior score according to some embodiments of this specification;

[0045] Figure 4 is an exemplary flow chart of determining a feature vector according to some embodiments of this specification;

[0046] Figure 5 is an exemplary schematic diagram of obtaining multi-source heterogeneous data according to some embodiments of this specification;

[0047] Figure 6 is an exemplary schematic diagram of determining failure risk according to some embodiments of this specification;

[0048] Figure 7 is a schematic structural diagram of a vehicle fault prediction device according to some embodiments of this specification;

[0049] Figure 8 This is a structural diagram of an electronic device according to some embodiments of this specification. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0051] In order to facilitate understanding of the implementation scheme provided in the examples of the present application, the relevant application background of the vehicle fault prediction method provided in the examples of the present application is first explained.

[0052] Currently, existing automotive fault diagnosis and prediction technologies typically rely on single data provided by the on-board diagnostic (OBD) system, which is relatively limited. These systems only generate fault codes after a fault occurs, which are used only for post-fault warnings and cannot predict potential risks in advance. Furthermore, the process by which a vehicle fails is often closely related to the driver's usage habits. For example, frequent sudden acceleration, prolonged high-speed operation, and frequent braking can accelerate component wear. Furthermore, the operating environment of a vehicle is complex and variable, with factors such as road conditions, temperature and humidity, and load also impacting the vehicle. However, most current systems fail to integrate multiple environmental data sources for joint modeling, resulting in poor adaptability of prediction algorithms and difficulty generalizing to different vehicles or scenarios. Consequently, they focus solely on single-dimensional data (such as OBD data) and lack comprehensive analysis of multi-dimensional data. This single-dimensional analysis approach makes it difficult to comprehensively assess a vehicle's operating status and fault risk.

[0053] In view of this, some embodiments of this specification provide a vehicle fault prediction method, which can provide richer information than a single data source by integrating device sensor data (such as OBD data), driving data (such as IMU data, GPS data, CAN data, etc.) and environmental data (such as temperature, humidity, road conditions), and can comprehensively evaluate the vehicle's operating status, thereby more accurately predicting potential faults.

[0054] Figure 1 This is an application scenario diagram of the vehicle fault prediction method shown in some embodiments of this specification.

[0055] The vehicle fault prediction method provided in the embodiments of the present application can be applied to the application scenario of vehicle fault monitoring, and can also be applied to other application scenarios, such as vehicle system testing, vehicle performance evaluation, etc.

[0056] The technical solution of the embodiments of the present application may be implemented as an electronic device, which may be deployed on a mobile device or connected to the mobile device via a wired or wireless connection. Alternatively, the electronic device may be the mobile device itself. The mobile device may have any appearance, such as an intelligent vehicle, an intelligent robot, or the like.

[0057] In some embodiments, the electronic device may be an onboard terminal integrated into a vehicle, such as an electronic control unit (ECU), a vehicle control unit (VCU), a microcontroller (MCU), or a device that interacts with the vehicle. The embodiments of the present application do not impose any restrictions on the specific type of electronic device.

[0058] When the electronic device is connected to a removable device, the electronic device can be a terminal device, such as a smart phone, tablet computer, laptop computer, desktop computer, etc., but not limited to this, or the electronic device can also be a server, which can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0059] In some embodiments, the application scenario may further include, for example, a network, a storage device, etc. The network may include any suitable wired or wireless network that can facilitate information and / or data exchange. The storage device may be used to store data, instructions, and / or any other information.

[0060] The following description will be made using the example of a vehicle as an example of a mobile device. It is worth noting that the application scenario of the vehicle fault prediction method is provided for illustrative purposes only and is not intended to limit the scope of this specification. For those skilled in the art, various changes and modifications can be made based on the description of this specification. For example, the application scenario may also include a database, an information source, etc. For another example, the application scenario may be implemented on other devices to achieve similar or different functions. However, these changes and modifications do not deviate from the scope of this specification.

[0061] Figure 2 This is an exemplary flow chart of a vehicle fault prediction method according to some embodiments of this specification. In some embodiments, the process can be executed based on an electronic device. Figure 2 As shown, the process includes the following steps.

[0062] Step 210 , obtaining multi-source heterogeneous data of the vehicle, where the multi-source heterogeneous data includes at least two types of vehicle device sensor data, driving data, and environmental data.

[0063] Multi-source heterogeneous data refers to data that differs in structure, format, or source. For example, multi-source heterogeneous data may include various data reflecting vehicle driving information.

[0064] Device sensor data is data used to monitor and reflect the operating status of various devices in the vehicle. For example, device sensor data may include on-board diagnostics (OBD) data. Exemplarily, device sensor data may include any one or a combination of engine speed, coolant temperature, fuel pressure, intake air temperature, battery voltage, engine load, and fault codes. Engine speed measures the number of engine revolutions per unit time (e.g., per minute). Coolant temperature measures the temperature of the engine coolant. Fuel pressure refers to the pressure in the fuel system. Intake air temperature measures the temperature of the engine's intake system. Battery voltage refers to the voltage of the vehicle's battery. Engine load measures the current engine load percentage. Diagnostic Trouble Codes (DTCs) are fault codes generated by the Electronic Control Unit (ECU) that indicate the location and cause of a vehicle fault.

[0065] In some embodiments, as Figure 5 As shown in the figure, various types of multi-source heterogeneous data can be obtained through environmental sensors, CAN bus, positioning system, IMU sensor, OBD interface, etc.

[0066] In some embodiments, the electronic device can connect to an OBD scanner or diagnostic tool via an OBD-II interface to periodically or in real time read the vehicle's device sensor data. Alternatively, the electronic device can obtain the vehicle's device sensor data periodically or in real time via a network connection based on an API interface provided by the vehicle manufacturer.

[0067] Driving data refers to data related to the driving behavior of the vehicle, including the vehicle's motion state, driving operations, etc. For example, driving data may include any one or a combination of vehicle speed, acceleration, angular velocity, brake pedal position, and accelerator pedal position. Vehicle speed is used to characterize the vehicle's driving speed. Acceleration is used to characterize the acceleration of the vehicle, including longitudinal acceleration (acceleration or deceleration) and lateral acceleration (lateral acceleration when turning). Angular velocity is used to characterize the angular velocity of the vehicle around the vertical axis. Brake pedal position may include the depth of the brake pedal being depressed. Accelerator pedal position may include the depth of the accelerator pedal being depressed.

[0068] In some embodiments, the electronic device can be connected to the inertial measurement unit (IMU), positioning system, CAN (Controller Area Network Bus) bus, etc. in the vehicle to obtain driving data under certain circumstances (for example, in real time, at intervals, or triggered under certain circumstances).

[0069] The positioning system may include a Global Positioning System (GPS) device, a Global Navigation Satellite System (GLONASS) device, a Compass Navigation System (COMPASS) device, a BeiDou Navigation Satellite System device, a Galileo Positioning System device, a Quasi-Zenith Satellite System (QZSS) device, etc. The positioning system may provide a real-time position and speed of a vehicle while it is traveling; for example, the position and speed of the vehicle at each point in time.

[0070] Environmental data refers to the external environmental conditions in which a vehicle operates. For example, environmental data may include weather data, road condition data (e.g., road slipperiness), and traffic conditions. Weather data may include information such as temperature, humidity, rainfall, and wind speed that characterizes environmental conditions. Road condition data may include road type (e.g., highways, urban roads, rural roads), road conditions (e.g., flatness, potholes, icing, etc.), traffic volume, and road slipperiness.

[0071] In some embodiments, the electronic device may be communicatively connected with an environmental sensor in the vehicle to obtain environmental data under certain circumstances (eg, in real time, at intervals, or triggered under certain circumstances).

[0072] Environmental sensors can include one or more sensors for acquiring environmental data. This data provides contextual information about driver behavior, helping to predict the potential impact of driving behavior on a malfunction. For example, cold weather can cause battery discharge, while slippery roads can increase brake wear.

[0073] In some embodiments, multi-source heterogeneous data may include large types of data such as device sensor data, driving data, and environmental data. Each large type of multi-source heterogeneous data may also include different small types of data.

[0074] In some embodiments, the device sensing data, driving data, environmental data, etc. are sequence data (eg, data within a period of time), and the device sensing data, driving data, environmental data are one-to-one corresponding in time.

[0075] The description of multi-source heterogeneous data is only given as an example, and it may also be in other forms or include other data.

[0076] Step 220 : extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors.

[0077] A feature vector is an array in which each element (feature) represents a characteristic of the data. For example, a feature vector for device sensor data might be [statistical value of X1, statistical value of X2, ...], a feature vector for driving data might be [statistical value of Y1, statistical value of Y2, ...], and a feature vector for environmental data might be [statistical value of Z1, statistical value of Z2, ...]. Here, X1 and X2 represent different types of device sensor data, Y1 and Y2 represent different types of driving data, and Z1 and Z2 represent different types of environmental data. Statistical values ​​can include one or a combination of the mean, standard deviation, maximum, and minimum values.

[0078] In some embodiments, feature extraction can be performed on different data in the multi-source heterogeneous data based on statistical methods (e.g., variance, mean, etc.), signal processing methods (e.g., filtering, frequency domain analysis, etc.), and machine learning methods (e.g., principal component analysis, etc.) to obtain multiple feature vectors. For example, feature vectors for certain multi-source heterogeneous data can be obtained based on the statistical features or time series features of the multi-source heterogeneous data.

[0079] In step 230 , based on the multiple feature vectors, a prediction model is used to obtain the failure risk of the vehicle in a future time period.

[0080] Failure risk refers to the likelihood that a vehicle will fail within a specific timeframe. Failure risk characterizes the likelihood that a component or assembly on a vehicle will fail. In some embodiments, failure risk can be expressed as a numerical value (e.g., failure severity, failure probability, etc.) or a level (e.g., failure grade).

[0081] The future time period is a period of time after the current time. The future time period can be the next week, the next month, the next three months, etc. For example, the prediction model can output that there is a 15% probability that the vehicle will experience an engine failure within the next week.

[0082] A predictive model is a model or algorithm used to determine the risk of a vehicle failing in a future time period.

[0083] In some embodiments, the prediction model is a machine learning model. For example, the prediction model may include any one or combination of a convolutional neural network (CNN), a recurrent neural network (RNN), a deep neural network (DNN), or other customized model structures.

[0084] In some embodiments, the input to the prediction model includes multiple feature vectors, and the output of the prediction model may include the vehicle's failure risk in a future time period. For example, multiple feature vectors may be concatenated and input into the prediction model. Exemplarily, the concatenated feature vector is X = [frequency of sudden acceleration, maximum steering angle, average battery voltage, number of fault codes, water temperature slope, average braking intensity, sudden braking frequency, standard deviation of idle vibration, short-term fuel correction value, weather temperature, road slip index, traffic flow, external temperature and humidity, etc.].

[0085] In some embodiments, the output of the prediction model may further include the fault type and the probability of the fault type.

[0086] In some embodiments, the prediction model can be trained based on a large number of first training samples with a first label through various feasible methods. For example, parameter updates can be performed based on a gradient descent method. An exemplary training process includes: inputting multiple labeled training samples into an initial prediction model, constructing a loss function based on the labels and the results of the initial prediction model, and iteratively updating the parameters of the initial prediction model based on the loss function through gradient descent or other methods. When preset conditions are met, model training is completed, and a trained prediction model is obtained. The preset conditions may include convergence of the loss function, the number of iterations reaching a threshold, etc.

[0087] In some embodiments, the first training sample includes at least a sample feature vector. The first training sample can be obtained based on historical data.

[0088] In some embodiments, the first label may include a failure risk corresponding to the first training sample. The first label may be obtained through automatic or manual labeling.

[0089] In some embodiments, the multi-source heterogeneous data may be sequence data. Sequence data refers to a series of data arranged in a specific order. For example, sequence data may refer to data arranged in chronological order. For another example, sequence data may also refer to data arranged in order of size, priority, etc.

[0090] In some embodiments, the sequence data may be a sequence consisting of sequence data of device sensing data, sequence data of driving data, sequence data of environmental data, and the like.

[0091] In some embodiments, in the sequence data of device sensor data, the sequence data of driving data, and the sequence data of environmental data, the time point corresponding to each data point of each type of sequence data can be the same as or similar to the time point corresponding to each data point in the other types of sequence data.

[0092] In some embodiments, the electronic device may periodically acquire sequence data of device sensor data, driving data, and environmental data within a period. Periodic acquisition refers to acquiring data continuously collected within a period at regular intervals. For example, the collected device sensor data, driving data, and environmental data may be sent to the electronic device every 10 seconds.

[0093] Continuous data collection refers to collecting data at consecutive time points. Continuous time points are multiple consecutive time points separated by a certain time interval. For example, the interval length can be any length, depending on actual needs. For example, the interval time can be 10ms, 20ms, etc.

[0094] In some embodiments, the prediction model is a time series-based machine learning model. A time series-based machine learning model may refer to a machine learning model that processes time series data. For example, a time series-based machine learning model may include a recurrent neural network (RNN), a bidirectional long short-term memory network (LSTM), a temporal convolutional network (TCN), and the like.

[0095] It is understandable that by analyzing sequence data through a time-series-based machine learning model, the risk of vehicle failure in future time periods can be better predicted and the accuracy of the output results can be improved.

[0096] In some embodiments, each type of multi-source heterogeneous data corresponds to a prediction model. For example, the feature vector corresponding to the device sensor data can be input into a first model to output a first failure risk for a future time period; the feature vector corresponding to the driving data can be input into a second model to output a second failure risk for a future time period; and the feature vector corresponding to the environmental data can be input into a third model to output a third failure risk for a future time period. Based on the first, second, and third failure risks, the vehicle's failure risk for the future time period is determined. For example, the first, second, and third failure risks can be determined based on a weighted sum or voting method. The first, second, and third models can be classification models that output failure risks such as failure type, failure probability, and failure location. For example, if the first model predicts a "high risk" for the vehicle's engine failure within the next week, the second model predicts a "low risk" for the vehicle's engine failure within the next week, and the third model predicts a "high risk" for the vehicle's engine failure within the next week, the final result is output based on statistical voting, with two votes for high risk and one for low risk, e.g., the vehicle's engine failure risk within the next week is "high risk."

[0097] In some embodiments, based on the output results of the prediction model, early warning information can be output to the user terminal. The early warning information is information related to the fault condition of the vehicle. The fault condition may include the fault type, fault location, fault probability, etc. The early warning information may be in the form of text, voice, image, video, tactile alarm, etc. or any combination thereof. The user terminal refers to one or more terminal devices or software used by the user. In some embodiments, the user terminal may be used by one or more users, including users who directly use the relevant programs for vehicle fault prediction, and may also include other relevant users. In some embodiments, the user terminal may be one or more devices with input and / or output functions, such as a mobile device, a tablet computer, a laptop computer, a desktop computer, or any combination thereof.

[0098] For example, the user terminal can be controlled to output warning information indicating fault risks in the form of pop-up windows or graphic prompts. Graphic warning information can include pictures, videos, etc., and can be displayed on the current interface of the user terminal. For example, different fault risks correspond to different fault levels and are displayed in different colors, such as green for low risk, yellow for medium risk, and red for high risk.

[0099] In some embodiments, as Figure 6 As shown, the relationship between the failure risk and the preset risk threshold can be determined. For example, when the failure risk is higher than the preset risk threshold, the user is reminded based on the early warning information; when the failure risk is lower than the preset risk threshold, the failure risk of the vehicle continues to be monitored through the above method.

[0100] In some embodiments, the prediction model can be deployed on a terminal device (such as an in-vehicle edge box, a local model, etc.). During actual use, the prediction model can be optimized, its configuration parameters updated, and fine-tuned to enhance its prediction capabilities and cope with new vehicle conditions or data structures. For example, the prediction model can be updated by pulling a remote configuration file at program startup or at a fixed period.

[0101] In some embodiments, the prediction model can be a lightweight model deployed in a vehicle-mounted terminal or edge computing device to achieve low-latency and high-real-time fault risk prediction.

[0102] In some embodiments of this specification, based on a variety of multi-source heterogeneous data, a prediction model is used to help efficiently and accurately predict the failure risk of a vehicle, thereby helping to improve the safety and reliability of the vehicle and reduce the probability of failure.

[0103] In some embodiments, the multi-source heterogeneous data includes first driving data, second driving data, and third driving data, and the plurality of feature vectors include driving behavior features.

[0104] Based on the method corresponding to each type, feature extraction is performed on different data in the multi-source heterogeneous data to obtain multiple feature vectors, including:

[0105] When the frequency of the first driving data being greater than or equal to the preset data threshold is greater than the preset frequency threshold, and / or

[0106] The duration during which the second driving data is greater than or equal to the preset data threshold is greater than the preset time threshold, and / or

[0107] When the third driving data is greater than or equal to the preset data threshold,

[0108] A driving behavior characteristic is determined based on the first driving data, the second driving data, and / or the third driving data.

[0109] Driving behavior features are features extracted from driving data that reflect driving behavior. For example, driving behavior features may include the frequency of sudden acceleration, the frequency of sudden braking, the duration of high speed, the duration of overspeeding, etc.

[0110] The first driving data, the second driving data, and the third driving data respectively refer to different types of driving data generated during the driving of the vehicle.

[0111] In some embodiments, the driving data may include multiple types, one type of driving data corresponds to one driving behavior feature, and the feature vector corresponding to the driving data may be a feature vector composed of driving behavior features corresponding to multiple types of driving data.

[0112] A preset data threshold is a pre-set value used to determine whether a specific driving data value meets or exceeds a certain numerical standard. Different types of driving data may correspond to different preset data thresholds, which are used to identify specific driving behaviors or vehicle states. For example, for the longitudinal acceleration of driving data, the corresponding preset data threshold is 2.5m / s², which is used to identify sudden acceleration behavior. For the angular velocity of driving data, the corresponding preset data threshold is 100° / s, which is used to identify sharp turns.

[0113] The preset frequency threshold is a pre-set value used to determine whether the frequency of a certain driving data reaches or exceeds a certain data standard. Different types of driving data may correspond to different preset frequency thresholds, which are used to identify the frequency of a specific driving behavior or vehicle state. For example, for the longitudinal acceleration of driving data, the corresponding preset data threshold is 2.5m / s 2 The preset frequency threshold is 5 times / minute. If the frequency of acceleration greater than or equal to 2.5m / s² exceeds 5 times / minute, frequent sudden acceleration is determined to exist, thereby obtaining the driving behavior characteristics corresponding to the driving data. For the angular velocity of the vehicle around the vertical axis in the driving data, the corresponding preset data threshold is 100° / s, and the preset frequency threshold is 3 times / minute. If the frequency of angular velocity greater than or equal to 100° / s exceeds 3 times / minute, frequent sharp turns are determined to exist, thereby obtaining the driving behavior characteristics corresponding to the driving data.

[0114] The preset time threshold is a pre-set time range used to determine whether a particular type of driving data meets or exceeds a certain time threshold. Different types of driving data may correspond to different preset time thresholds, which are used to identify the duration of a specific driving behavior or vehicle state. For example, for driving data with speed, the corresponding preset time threshold is 15 minutes. If the vehicle speed is greater than or equal to 120 km / h for more than 15 minutes, it is determined to be a prolonged period of speeding, thus obtaining the driving behavior characteristics corresponding to the driving data.

[0115] The preset data threshold, preset frequency threshold, preset time threshold system default value, experience value, manually preset value, etc. or any combination thereof can be set according to actual needs, and this manual does not limit this.

[0116] It can be understood that when the frequency of a certain driving data being greater than or equal to a preset data threshold is greater than the preset frequency threshold, and / or the duration being greater than or equal to the preset data threshold is greater than the preset time threshold, and / or is greater than or equal to the preset data threshold, the driving behavior feature can be any one or combination of the driving data itself, or the frequency corresponding to the driving data, or the duration corresponding to the driving data, etc.

[0117] In some embodiments of this specification, by integrating the first, second, and third driving data, driving behavior can be evaluated from multiple dimensions. For example, acceleration data can reflect sudden acceleration, angular velocity data can reflect sharp turns, and brake pedal position data can reflect sudden braking. Comprehensive analysis of multiple driving data types allows for a more comprehensive assessment of the impact of driving behavior on the vehicle. Compared to relying on a single data type, this can improve the accuracy of vehicle fault prediction, thereby identifying potential fault risks in advance and providing early warnings, helping users and maintenance personnel take preemptive measures to reduce the probability of faults.

[0118] In some embodiments, the first driving data includes inertial motion data of the vehicle, and the driving behavior characteristics corresponding to the first driving data include at least one of acceleration information, start-stop information, angular velocity information, and / or

[0119] The second driving data includes vehicle positioning data, and the driving behavior characteristics corresponding to the second driving data include speeding information, and / or

[0120] The third driving data includes bus data of the vehicle, and the driving behavior characteristics corresponding to the third driving data include starting vibration information and / or braking intensity information.

[0121] Inertial motion data refers to the data provided by the vehicle's inertial measurement unit sensor, which is used to describe the vehicle's motion state.

[0122] In some embodiments, the inertial measurement unit may include an inertial sensor. For example, the inertial sensor may include a motion sensor (e.g., an accelerometer) and a rotation sensor (e.g., a gyroscope). In some embodiments, the inertial sensor may be supplemented by a barometric altimeter. In some embodiments, the inertial sensor may be supplemented by a magnetic sensor (e.g., a magnetometer) and / or a velocity measurement device.

[0123] In some embodiments, the inertial measurement unit includes three single-axis accelerometers and three single-axis gyroscopes. The accelerometers are used to detect the vehicle's acceleration signals along the three axes of the vehicle coordinate system, while the gyroscopes are used to detect the vehicle's angular velocity signals. In other words, the inertial motion data can include information such as the vehicle's acceleration signals along the three axes, its angular velocity along the three axes, and its direction.

[0124] Acceleration information is information related to the acceleration behavior of the vehicle. For example, acceleration information may include the frequency of rapid acceleration and the duration of rapid acceleration.

[0125] The frequency of rapid acceleration refers to the acceleration per unit time greater than or equal to the preset data threshold (such as 2.5m / s 2The unit time can be a time period such as one second, one minute, or one hour. The acceleration duration refers to the duration during which each acceleration is greater than or equal to a preset data threshold.

[0126] For example, the electronic device can extract the acceleration of the vehicle in the driving direction from the IMU data (i.e., inertial motion data) and the accelerator pedal position from the CAN data (i.e., bus data) to determine whether the accelerator pedal is pressed: when the accelerator pedal is pressed and the acceleration is greater than 2.5m / s 2 It is counted as a sudden acceleration, and the number of sudden accelerations in unit time is counted to obtain the driving behavior characteristics corresponding to the driving data.

[0127] Start-stop information is information related to the vehicle's start-stop behavior. For example, start-stop information can include the frequency of frequent starts and stops, the duration of starts and stops, etc.

[0128] Frequent start-stop frequency refers to the number of times a vehicle switches from stationary to moving and then back to stationary, within a unit of time, that exceeds a preset frequency threshold. Start-stop duration refers to the duration of each stop-start, start-stop, and back to stationary state.

[0129] The start-up vibration information is information related to the vehicle's starting behavior. For example, the start-up vibration information may include the start-up vibration intensity, the start-up vibration duration, etc.

[0130] The starting vibration intensity refers to the peak acceleration of the vehicle when starting, and the starting vibration duration refers to the duration of the vehicle's vibration when starting.

[0131] For example, the electronic device may obtain acceleration from the accelerometer of the IMU, and calculate the value or duration of the acceleration exceeding a preset data threshold to obtain the driving behavior characteristics corresponding to the driving data.

[0132] Angular velocity information is information related to the vehicle's turning behavior. For example, the angular velocity information may include the frequency of sharp turns, the duration of sharp turns, and the angular velocity of high-speed sharp turns.

[0133] Sharp turn frequency refers to the number of times per unit time that the angular velocity is greater than or equal to a preset threshold (e.g., 100° / s). Angular velocity duration refers to the duration of each period in which the angular velocity is greater than or equal to the preset threshold. Sharp turn angular velocity refers to the angular velocity of a vehicle while speeding.

[0134] For example, the electronic device can obtain the angular velocity of the vehicle around the vertical axis through the gyroscope of the IMU, calculate the absolute value of the angular velocity, record the angular velocity value when the absolute value of the angular velocity is greater than 100° / s, and count the number of times the absolute value of the angular velocity of the vehicle is greater than 100° / s or the average angular velocity when the vehicle is speeding, so as to obtain the driving behavior characteristics corresponding to the driving data.

[0135] In some embodiments, the electronic device can obtain acceleration from the accelerometer of the IMU, and obtain the brake pedal position and the accelerator pedal position from the CAN bus. When the acceleration suddenly changes from a positive value to a negative value, and the accelerator pedal changes from being pressed to being released, and the brake pedal changes from being released to being pressed, it is counted as a start and stop. The number of starts and stops in unit time, or the frequency or average interval time of vehicle starts and stops, is counted to obtain the driving behavior characteristics corresponding to the driving data.

[0136] By identifying driving behavior characteristics, the impact of driving behavior on vehicle load and wear can be analyzed, helping to identify driving behaviors that may lead to failure risks. For example, frequent hard braking may cause premature wear of the brake system.

[0137] Positioning data refers to data provided by a vehicle's positioning system (such as GPS) that describes the vehicle's geographic location and driving trajectory. For example, positioning data typically includes latitude and longitude, speed, etc.

[0138] Speeding information is information related to the speeding behavior of the vehicle. For example, speeding information may include speeding frequency, speeding duration, speed, etc.

[0139] The speeding frequency refers to the number of times the vehicle's speed is greater than or equal to a preset data threshold (e.g., 120 km / h) per unit time. The speeding duration refers to the duration of each time the vehicle's speed is greater than or equal to the preset data threshold.

[0140] For example, the vehicle's driving speed can be obtained from the positioning data, the speeding time in which the driving speed exceeds a preset data threshold in a unit time can be counted, and the proportion of the speeding time to the total driving time can be calculated to obtain the speeding ratio, so as to obtain the driving behavior characteristics corresponding to the driving data.

[0141] Bus data refers to data transmitted via a vehicle's communication bus (such as the CAN bus) and is used to describe communication between various electronic control units (ECUs) within the vehicle. Bus data can include engine status information, brake system status information, instrument panel information, and more.

[0142] Bus data can transmit the vehicle's operating status and operation commands in real time, helping to determine whether the vehicle has abnormalities or failure risks.

[0143] Braking intensity information is information related to the vehicle's braking behavior. For example, braking intensity information may include the frequency of sudden braking, brake pedal position, etc.

[0144] The frequency of emergency braking refers to the deceleration per unit time being less than or equal to the preset data threshold (such as -2.5m / s 2 ) times. Brake pedal position refers to the depth of the brake pedal depression.

[0145] For example, the electronic device may obtain the brake pedal position from the bus data, and calculate the average or maximum value of the brake pedal depth to obtain the driving behavior characteristics corresponding to the driving data.

[0146] In some embodiments of the present specification, frequent sudden acceleration and braking may cause excessive wear on the vehicle's braking system and engine. By monitoring these behaviors, potential faults can be predicted in advance; by analyzing the accelerator pedal position and brake pedal position data, the impact of driving behavior on the vehicle's health can be evaluated; by analyzing the vehicle speed data, the impact of speeding behavior on the vehicle's health can be evaluated, which helps to comprehensively evaluate the impact of driving behavior on the vehicle's health, and then be used for fault prediction, so that potential faults can be predicted in advance.

[0147] In some embodiments, the device sensor data includes at least one of battery data, engine data, fault code number, and fuel tank data, and the plurality of feature vectors include fault features.

[0148] Based on the method corresponding to each type, feature extraction is performed on different data in the multi-source heterogeneous data to obtain multiple feature vectors, including:

[0149] When the number of fault codes is greater than or equal to a preset number threshold; and / or

[0150] When the battery data is abnormal; and / or the engine data is abnormal; and / or the fuel tank data is abnormal;

[0151] A fault characteristic is determined based on at least one of the fault code number, battery data, engine data, and fuel tank data.

[0152] In some embodiments, the device sensor data may include multiple types, one type of device sensor data corresponds to one fault feature, and the feature vector corresponding to the device sensor data may be a feature vector composed of fault features corresponding to multiple types of device sensor data.

[0153] Battery data refers to the operating status data of the vehicle battery. For example, battery data may include battery voltage, battery current, battery temperature, etc.

[0154] In some embodiments, the abnormality in the battery data may include abnormal battery voltage, abnormal battery current, abnormal battery temperature, etc.

[0155] The battery voltage abnormality may be that the battery voltage is lower or higher than a preset voltage range (such as lower than 9V or higher than 14.5V).

[0156] Battery current abnormality may be that the battery charging current or discharging current is lower or higher than the preset current range.

[0157] Abnormal battery temperature may be that the battery temperature is lower or higher than a preset temperature range.

[0158] Engine data refers to engine operating status data. For example, engine data may include engine speed, coolant temperature, fuel pressure, intake pressure, etc.

[0159] In some embodiments, the abnormalities in the engine data may include abnormal engine speed, abnormal coolant temperature, abnormal fuel pressure, abnormal intake pressure, etc.

[0160] The abnormal engine speed may be that the engine speed is lower than or higher than a preset speed range.

[0161] The abnormal coolant temperature may be that the coolant temperature is lower than or higher than a preset cooling temperature range.

[0162] Fuel pressure abnormality may be that the fuel pressure is lower or higher than a preset fuel pressure range.

[0163] The intake pressure abnormality may be that the intake pressure is lower than or higher than a preset intake pressure range.

[0164] Fuel tank data refers to the operating status data of the fuel tank, for example, the fuel tank data may include fuel level, fuel temperature, etc.

[0165] In some embodiments, the abnormalities in the fuel tank data may include abnormal fuel level, abnormal fuel temperature, etc.

[0166] The fuel level abnormality may be that the fuel level is lower or higher than the preset fuel level range.

[0167] The fuel temperature abnormality may be that the fuel temperature is lower than or higher than a preset fuel temperature range.

[0168] The preset ranges corresponding to various data may be determined based on experiments or experience.

[0169] In some embodiments, the fault signature may be determined in a variety of ways.

[0170] For example, the electronic device can read the number of fault codes through OBD PID 01. If the number of fault codes is greater than 0, it means that there is a confirmed fault. The more fault codes there are, the more complex the problem may be.

[0171] For example, the electronic device can read the minimum battery voltage through the OBD PID 42. If the minimum battery voltage is lower than a preset data threshold (such as 9V), it indicates that the battery may be aged or there may be a problem with the charging system.

[0172] For example, the electronic device can read the short-term fuel correction value through OBD PID 06. If the short-term fuel correction value continuously exceeds the preset data range (such as -20% to +20%), it indicates that there may be a problem with the fuel system, such as abnormal fuel pressure or air leakage in the intake system.

[0173] For example, the electronic device can read the temperature change trend through OBD PID 05. The temperature change trend is used to record the changes in the engine temperature from startup to normal operating temperature. If the temperature rise curve is abnormally slow or fluctuates, it means that there may be a problem with the cooling system, such as insufficient coolant or a clogged radiator.

[0174] For example, the electronic device can read the coolant temperature through OBD PID 05. If the coolant temperature is continuously abnormal (such as too high or too low), it may indicate a cooling system failure or a sensor damage.

[0175] For example, the electronic device can read the intake air temperature through the OBD PID 0F. If the intake air temperature sensor data is abnormal (such as too high or too low), it means that there may be a sensor failure or a problem with the intake system.

[0176] For example, the electronic device can read the fuel pressure sensor data through OBD PID 0A. If the fuel pressure sensor data is abnormal (such as the fuel pressure is too high or too low), it indicates that there may be a problem with the fuel system.

[0177] For example, the electronic device can read the oxygen sensor data through the OBD PID 14. If the oxygen sensor data is abnormal (such as the sensor voltage is too high or too low), it may indicate a sensor failure or an abnormal fuel mixture ratio.

[0178] For example, the electronic device can read the throttle position sensor data through the OBD PID 11. If the throttle position sensor data is abnormal (such as the sensor voltage is too high or too low), it may indicate a sensor failure or a stuck throttle.

[0179] In some embodiments of this specification, by integrating battery data, engine data, fault code counts, and fuel tank data, the health status of the vehicle can be evaluated from multiple dimensions. The evaluation is more comprehensive than that of a single data type, and can more accurately identify potential fault problems. Potential fault risks can be identified in advance, providing early warnings to help vehicle owners and maintenance personnel take measures in advance and reduce the probability of failures.

[0180] Figure 3 This is an exemplary flow chart of determining a driving behavior score according to some embodiments of this specification. In some embodiments, the process can be executed based on an electronic device. Figure 3 As shown, the process includes the following steps.

[0181] In some embodiments, the feature vector also includes environmental features, and the input of the prediction model also includes a driving behavior score, which is determined by:

[0182] Step 310 : constructing a cause-effect relationship graph based on driving behavior characteristics, fault characteristics, and environmental characteristics.

[0183] The causal relationship graph is a graph-structured data consisting of nodes and edges. A node corresponds to a type of multi-source heterogeneous data, and the node's features include the feature vector corresponding to the multi-source heterogeneous data. An edge is related to the causal relationship of the fault, and the edge's features include the causal strength of the causal relationship.

[0184] Step 320 : Determine the driving behavior score based on the cause-effect relationship diagram and a preset scoring model.

[0185] The driving behavior score is used to quantify the degree to which driving behavior characteristics affect vehicle health.

[0186] In some embodiments, multiple driving behavior characteristics may be scored based on expert experience or other methods, and the scores of the multiple driving behaviors may be weighted and summed to obtain a total driving behavior score, which is then input into a prediction model for fault prediction.

[0187] Causal graphs are used to represent the causal relationships between different nodes. Graph-structured data consists of nodes and edges. Nodes can include features, such as driving behavior features corresponding to driving data, fault features corresponding to device sensor data, or environmental features corresponding to environmental data. Node data can come from different sensors or systems. For example, device sensor data, driving data, and environmental data. Nodes can be numerical (such as temperature and vehicle speed) or categorical (such as weather type and road conditions).

[0188] Each edge represents a causal relationship between two nodes. In some embodiments, edges connect nodes, and nodes have corresponding node characteristics and edges have corresponding edge characteristics. For example, nodes may include nodes for driving behavior characteristics, nodes for fault characteristics, and nodes for environmental characteristics. Among them, the characteristics of the nodes for driving behavior characteristics may include the frequency of sudden acceleration, the frequency of sudden braking, the duration of speeding, etc. The characteristics of the nodes for fault characteristics may include the number of fault codes, abnormal battery voltage, abnormal engine speed, etc. The characteristics of the nodes for environmental characteristics may include weather conditions, road conditions, traffic flow, etc.

[0189] An edge represents the causal relationship between two nodes, that is, the degree to which the state of one node affects the state of another node.

[0190] Edge features describe the properties of the causal relationship between nodes. The features of each edge include weight, causal direction, and so on. Causal strength can be a numerical value that indicates the strength of the causal relationship. For example, causal strength can be a value between 0 and 1, with 1 indicating a strong causal relationship and 0 indicating no causal relationship. Causal direction indicates the direction of the causal relationship. For example, from the driver's sudden acceleration behavior to the increase in engine load, and then to the increase in engine temperature. The weight of the edge indicates the causal strength of the causal relationship. For example, frequent sudden braking behavior may have a greater impact on the wear of the brake system, while less sudden braking behavior may have a smaller impact on the system.

[0191] In some embodiments, the causal relationship diagram includes the path to the fault. For example, the path to fault code P0128 includes: high-frequency rapid acceleration + high-temperature operation → increased engine load → increased cooling system load → abnormal water temperature → fault code P0128 appears. Another example is the path of environmental data leading to a fault, including: low temperature → increased battery discharge rate → decreased battery voltage → increased battery aging risk. Another example is the path to a brake system fault, including: high braking frequency → increased brake pad temperature → increased brake system wear → brake system failure.

[0192] Node and edge characteristics can be determined using various methods based on underlying data. The data source can be the methods described in other embodiments or other methods. The data can include current data as well as historical data. For example, device sensor data, driving data, and environmental data can be acquired using sensors. An electronic device can sequence data from multiple sensors, which can include sensor type, sensor location, sensor reading, and so on.

[0193] In some embodiments, the electronic device can use a preset scoring model to predict sub-scores for each driving behavior characteristic in the current device sensor data, driving data, and environmental data based on the cause-and-effect relationship graph. The driving behavior score can be determined by weighted summation of the sub-scores and their corresponding weights. The weights can be determined based on experiments or experience.

[0194] In some embodiments, the preset scoring model can be a graph neural network model (GNN) or other graph models, such as a graph convolutional neural network model (GCNN), or other processing layers can be added to the graph neural network model, its processing method can be modified, etc.

[0195] In some embodiments, the preset scoring model can be obtained by training based on training data. The training data includes a second training sample and a second label. For example, the second training sample may include a historical causal relationship graph, and the second label may be a driving behavior score corresponding to each node of a driving behavior feature in the historical causal relationship graph. The second training sample may be determined based on historical data, and the second label may be determined by electronic equipment or human annotation.

[0196] In some embodiments of the present specification, the causal relationship between driving behavior and vehicle health can be determined in a preset scoring model through a causal relationship diagram, such as the impact of timely behaviors such as sudden acceleration and sudden braking on the vehicle's health status (such as engine temperature, battery voltage, etc.), thereby evaluating the impact of driving behavior on the risk of vehicle failure and accurately predicting the future risk of vehicle failure.

[0197] Figure 4 This is an exemplary flow chart of determining a feature vector according to some embodiments of this specification. In some embodiments, this process can be executed based on an electronic device. Figure 4 As shown, the process includes the following steps.

[0198] In some embodiments, feature extraction is performed on different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors, including:

[0199] Step 410: preprocess the multi-source heterogeneous data based on the collection characteristics of the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data;

[0200] In step 420 , feature extraction is performed on the various pre-processed multi-source heterogeneous data to obtain multiple feature vectors.

[0201] Collection characteristics refer to information related to the data collection process, etc. For example, collection characteristics may include the timestamp of collection, collection frequency, data source, etc.

[0202] Preprocessing may include but is not limited to data cleaning, data conversion, data normalization, data filtering (eg, filtering out high-frequency noise through median filtering, low-pass filtering), etc.

[0203] In some embodiments, multi-source heterogeneous data with different collection characteristics adopt different preprocessing methods.

[0204] In some embodiments, the acquisition feature includes a sampling frequency. Based on the acquisition feature of the multi-source heterogeneous data, the multi-source heterogeneous data is preprocessed to obtain preprocessed multi-source heterogeneous data, including:

[0205] Based on the first multi-source heterogeneous data with the highest sampling frequency, the second multi-source heterogeneous data is resampled to obtain the resampled second multi-source heterogeneous data that is aligned with the first multi-source heterogeneous data at multiple time points, so as to obtain the preprocessed multi-source heterogeneous data, wherein the second multi-source heterogeneous data is the multi-source heterogeneous data other than the first multi-source heterogeneous data.

[0206] The sampling frequency refers to the number of times data is collected per unit time. For example, if the sampling frequency of multi-source heterogeneous data is 10 Hz, it means that 10 data points are collected per second.

[0207] The first multi-source heterogeneous data refers to data collected by a data source with the highest sampling frequency among the multi-source heterogeneous data.

[0208] Resampling is the process of resampling data to adjust the sampling frequency of the data to match the target sampling frequency (e.g., the highest sampling frequency). Resampling includes upsampling (increasing the sampling frequency) and downsampling (decreasing the sampling frequency).

[0209] Upsampling refers to adjusting multi-source heterogeneous data with a lower sampling frequency to multi-source heterogeneous data with a higher sampling frequency. Upsampling can be achieved by inserting new data points between two data points using interpolation methods (such as linear interpolation and spline interpolation).

[0210] Downsampling refers to adjusting multi-source heterogeneous data with a higher sampling frequency to multi-source heterogeneous data with a lower sampling frequency. Downsampling can be achieved by taking one data point every other data point.

[0211] For example, the first data multi-source heterogeneous data consists of data points at time points t1, t2, t3, ..., then the data points in the resampled second multi-source heterogeneous data correspond one-to-one to the data points in the first data multi-source heterogeneous data at time points t1, t2, t3, ...

[0212] In some embodiments of this specification, by preprocessing multi-source heterogeneous data for time alignment, selecting the data source with the highest sampling frequency as a benchmark, and resampling other data sources to align all data in time, the consistency of multi-source heterogeneous data with different sampling frequencies in the time dimension can be ensured, facilitating subsequent fault prediction analysis.

[0213] In some embodiments, feature extraction is performed on various pre-processed multi-source heterogeneous data to obtain multiple feature vectors, including:

[0214] A sliding window based on a preset time length slides on the preprocessed multi-source heterogeneous data according to a preset step length to divide the preprocessed multi-source heterogeneous data into multiple interval data;

[0215] Feature extraction is performed on the statistical features of multiple interval data to obtain the feature vectors corresponding to multi-source heterogeneous data.

[0216] A sliding window can refer to a window that partitions multi-source heterogeneous data by sliding. The sliding window length can refer to the amount of multi-source heterogeneous data contained within the sliding window or the time duration of the sliding window, i.e., the length of the sliding window. The preset step size can be the amount of data between adjacent sliding windows. For example, the sliding window length can be 1 second, 3 seconds, 5 seconds, etc.

[0217] For example, the sampling frequency of device sensor data is once every 1 second; the sampling frequency of inertial motion data is once every 0.1 second; and the sampling frequency of positioning data is once every 2 seconds. The high-frequency signal (such as inertial motion data) can be used as the main time axis to interpolate / align the low-frequency signal, and all multi-source heterogeneous data can be slidingly combined according to a sliding time window (such as every 1 second).

[0218] Interval data refers to a collection of multiple data points obtained by partitioning multi-source heterogeneous data according to a sliding window length and a preset step size. In some embodiments, the corresponding data can be obtained by partitioning multi-source heterogeneous data by data volume or time. In some embodiments, the sliding window can be a time interval determined by the starting and ending data positions.

[0219] In some embodiments, the electronic device can slide the sliding window from front to back based on the order of data points in the multi-source heterogeneous data. Each time the sliding window is slid according to a preset step size, a segment of data corresponding to the sliding window length is intercepted and determined as interval data. This is repeated until all multi-source heterogeneous data has been intercepted. For example, assuming that the multi-source heterogeneous data includes data 1, data 2, ..., data n, the sliding window length is 3, and the preset step size is 2, the following multiple interval data can be obtained: (data 1, data 2, data 3), (data 3, data 4, data 5), ​​..., (data n-2, data n-1, data n).

[0220] In some embodiments, the sliding window length and the preset step size are determined based on experiments or experience.

[0221] In some embodiments, statistical features refer to descriptive statistics calculated from the data, such as mean, variance, maximum, minimum, median, etc. For each interval data point, these statistical features are calculated and combined into a vector. For example, if the statistical features include mean, variance, and maximum, the vector for each interval data point can be represented as follows: feature vector of interval data = [mean, variance, maximum], etc.

[0222] In some embodiments, for each type of multi-source heterogeneous data (such as driving data (bus data, inertial motion data, positioning data), equipment sensor data (battery data, engine data, number of fault codes, fuel tank data), environmental data, etc.), a similar method as described above can be used to obtain multiple interval data corresponding to the multi-source heterogeneous data, perform statistical analysis on the multiple interval data corresponding to the multi-source heterogeneous data, obtain statistical features corresponding to the multiple interval data, combine the multiple statistical features to obtain statistically analyzed multi-source heterogeneous data, and obtain the feature vector of the multi-source heterogeneous data through the corresponding feature extraction method for the statistically analyzed multi-source heterogeneous data.

[0223] In some embodiments of the present specification, by dividing multi-source heterogeneous data into multiple sub-data segments, continuous time series data can be decomposed into multiple segments of fixed length, thereby reducing the overall complexity of the data.

[0224] It should be noted that the above description of the relevant processes is for illustration and purpose only and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the processes under the guidance of this specification. However, such modifications and changes are still within the scope of this specification.

[0225] Figure 7 It is a structural schematic diagram of a vehicle fault prediction device according to some embodiments of this specification.

[0226] like Figure 7 As shown, one or more embodiments of this specification also provide a structural schematic diagram of a vehicle fault prediction device. The vehicle fault prediction device may include:

[0227] An acquisition module 701 is configured to acquire multi-source heterogeneous data of a vehicle, where the multi-source heterogeneous data includes at least two types of vehicle device sensor data, driving data, and environmental data;

[0228] Extraction module 702, configured to extract features from different data in the multi-source heterogeneous data based on a method corresponding to each type, to obtain multiple feature vectors;

[0229] The prediction module 703 is used to obtain the failure risk of the vehicle in the future time period based on multiple feature vectors through a prediction model.

[0230] Among them, the acquisition module 701, the extraction module 702, and the prediction module 703 can be used to execute the corresponding embodiments of the above-mentioned vehicle fault prediction method respectively. For the specific implementation methods of these modules and more details, please refer to the corresponding method part, which will not be repeated here.

[0231] The specific implementation of the above operations can be found in the previous embodiments and will not be repeated here.

[0232] Figure 8 This is a structural diagram of an electronic device according to some embodiments of this specification.

[0233] The embodiment of the present application further provides an electronic device 800, which may include one or more processors 801 of processing cores, one or more computer-readable storage media memories 802, a power supply 803, an input unit 804, and other components. Those skilled in the art will appreciate that Figure 8 The electronic device structure shown in the figure does not constitute a limitation of the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently.

[0234] Processor 801 is the vehicle's fault prediction center, connecting the various components of the entire electronic device using various interfaces and lines. By running or executing software programs and / or modules stored in memory 802 and accessing data stored in memory 802, it performs various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. It is understood that processor 801 transmits signals with the controller. Optionally, processor 801 may include one or more processing cores; preferably, processor 801 may integrate an application processor and a modem processor, wherein the application processor primarily handles the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into processor 801.

[0235] The memory 802 can be used to store software programs and modules. The processor 801 executes various functional applications and data processing by running the software programs and modules stored in the memory 802. The memory 802 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 802 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 802 may also include a memory controller to provide the processor 801 with access to the memory 802.

[0236] In some embodiments of the present application, the vehicle fault prediction device can be implemented in the form of a computer program. The computer program can be used in Figure 8 The electronic device is operated on the electronic device shown. The memory of the electronic device can store the various program modules that constitute the vehicle fault prediction device. The computer program composed of the various program modules enables the processor to execute the steps of the vehicle fault prediction method of each embodiment of the present application described in this specification.

[0237] The electronic device includes a processor, memory, and a network interface connected via a system bus. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the electronic device is configured to communicate with external electronic devices via a network connection. When executed by the processor, the computer program implements a vehicle fault prediction method.

[0238] The electronic device also includes a power supply 803 for supplying power to various components. Preferably, the power supply 803 can be logically connected to the processor 801 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The power supply 803 can also include one or more DC or AC power supplies, a recharging system, a power failure detection circuit, a power converter or inverter, a power status indicator, and other arbitrary components.

[0239] The electronic device may further include an input unit 804, which may be configured to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0240] Although not shown, the electronic device may further include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 801 in the electronic device loads the executable files corresponding to one or more application processes into the memory 802 according to computer instructions, and the processor 801 runs the application stored in the memory 802 to implement various functions, such as the vehicle fault prediction method of each embodiment of the present application described in this specification.

[0241] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be accomplished by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0242] In specific implementation, the above units or structures can be implemented as independent entities, or can be arbitrarily combined to implement as the same or several entities. The specific implementation of the above units or structures can refer to the previous method embodiments and will not be repeated here.

[0243] It should be noted that Figure 8 This is only one implementation of the electronic device 800 provided in the embodiment of the present application. In actual applications, the electronic device 800 may also include more or fewer components, which is not limited here.

[0244] It should be understood that the various schemes of the embodiments of the present application can be reasonably combined and used, and the explanations or descriptions of the various terms appearing in the embodiments can be referenced or explained with each other in the various embodiments, without limitation to this.

[0245] It should also be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0246] Based on the above embodiments and the same concept, an embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a computer, the computer executes the method provided in the above embodiments.

[0247] Based on the above embodiments and the same concept, an embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program runs on a computer, it enables the computer to execute the method provided in the above embodiments.

[0248] The present application also provides a vehicle comprising the vehicle fault prediction device described in any embodiment; or comprising the electronic device described in any embodiment. The vehicle may be a gasoline-powered vehicle, a plug-in hybrid vehicle, or a new energy vehicle, and this specification does not specifically limit this.

[0249] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0250] The embodiments, implementation methods and related technical features of the present application can be combined and replaced with each other without conflict.

[0251] The above are only preferred embodiments of the present application and do not constitute any form of limitation to the present application. Although the descriptions of each embodiment in the embodiments of the present application have different focuses, for parts that are not described in detail in a certain embodiment, please refer to the relevant embodiments of other embodiments. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application are still within the scope of the technical solution of the present application.

Claims

1. A vehicle fault prediction method, characterized in that: The method comprises: Acquiring multi-source heterogeneous data of a vehicle, wherein the multi-source heterogeneous data includes at least two types of data selected from the group consisting of device sensor data, driving data, and environmental data of the vehicle; Extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each of the types to obtain multiple feature vectors; Based on the multiple feature vectors, a prediction model is used to obtain the failure risk of the vehicle in a future time period.

2. The method according to claim 1, characterized in that The multi-source heterogeneous data includes first driving data, second driving data and third driving data, the multiple feature vectors include driving behavior features, The method of extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes: When the frequency of the first driving data being greater than or equal to the preset data threshold is greater than the preset frequency threshold, and / or The duration of the second driving data being greater than or equal to the preset data threshold is greater than a preset time threshold, and / or When the third driving data is greater than or equal to the preset data threshold, The driving behavior characteristic is determined based on the first driving data, the second driving data and / or the third driving data.

3. The method according to claim 2, characterized in that The first driving data includes inertial motion data of the vehicle, and the driving behavior characteristics corresponding to the first driving data include at least one of acceleration information, start-stop information, and angular velocity information, and / or The second driving data includes positioning data of the vehicle, and the driving behavior characteristics corresponding to the second driving data include speeding information, and / or The third driving data includes bus data of the vehicle, and the driving behavior characteristics corresponding to the third driving data include starting vibration information and / or braking intensity information.

4. The method according to claim 1, wherein The device sensor data includes at least one of battery data, engine data, fault code number, and fuel tank data, and the multiple feature vectors include fault features. The method of extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes: When the number of fault codes is greater than or equal to a preset number threshold; and / or When the battery data is abnormal; and / or the engine data is abnormal; and / or the fuel tank data is abnormal; The fault characteristic is determined based on at least one of the fault code number, the battery data, the engine data, and the fuel tank data.

5. The method according to claim 3 or 4, characterized in that The feature vector also includes environmental features, and the input of the prediction model also includes a driving behavior score, which is determined by: A causal relationship graph is constructed based on the driving behavior characteristics, the fault characteristics, and the environmental characteristics; the causal relationship graph is graph structure data, and the graph structure data includes nodes and edges; each node corresponds to a type of the multi-source heterogeneous data, and the characteristics of the node include a feature vector corresponding to the multi-source heterogeneous data; the edges are related to the causal relationship of the fault occurrence, and the characteristics of the edges include the causal strength of the causal relationship; Based on the cause-effect relationship diagram, the driving behavior score is determined by a preset scoring model.

6. The method according to claim 1, characterized in that The method of extracting features from different data in the multi-source heterogeneous data based on a method corresponding to each type to obtain multiple feature vectors includes: Preprocessing the multi-source heterogeneous data based on the collection characteristics of the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; Feature extraction is performed on the plurality of preprocessed multi-source heterogeneous data to obtain the plurality of feature vectors.

7. The method according to claim 6, characterized in that The acquisition feature includes a sampling frequency. The preprocessing of the multi-source heterogeneous data based on the acquisition feature of the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data includes: Based on the first multi-source heterogeneous data with the highest sampling frequency, the second multi-source heterogeneous data is resampled to obtain the resampled second multi-source heterogeneous data that is aligned with the first multi-source heterogeneous data at multiple time points, so as to obtain the preprocessed multi-source heterogeneous data, wherein the second multi-source heterogeneous data is the multi-source heterogeneous data other than the first multi-source heterogeneous data.

8. The method according to claim 6, characterized in that The step of extracting features from the plurality of pre-processed multi-source heterogeneous data to obtain the plurality of feature vectors comprises: Sliding a sliding window based on a preset time length on the pre-processed multi-source heterogeneous data according to a preset step length to divide the pre-processed multi-source heterogeneous data into a plurality of interval data; Feature extraction is performed on the statistical features of the multiple interval data to obtain feature vectors corresponding to the multi-source heterogeneous data.

9. A vehicle fault prediction device, characterized in that: The device comprises: an acquisition module, configured to acquire multi-source heterogeneous data of a vehicle, wherein the multi-source heterogeneous data includes at least two types of data selected from the group consisting of device sensor data, driving data, and environmental data of the vehicle; an extraction module, configured to extract features from different data in the multi-source heterogeneous data based on a method corresponding to each of the types, to obtain a plurality of feature vectors; The prediction module is used to obtain the failure risk of the vehicle in a future time period based on the multiple feature vectors through a prediction model.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images