Driving data processing method, monitoring device, storage medium and controller

Abstract

The present invention belongs to the field of fault diagnosis technology, and in particular relates to a driving data processing method, monitoring device, storage medium and controller; historical data of multiple faulty vehicles are collected through the cloud and window samples are constructed by segmenting the historical data of each vehicle, and then fault prediction of a target component is performed under the drive of the window data; wherein, the target component can be an important component that sends vehicle control information, i.e., an accelerator pedal; further, a fault model of the target component is constructed by feature extraction of the window data and pattern recognition or machine learning of cloud vehicle data, and the driving data updated in real time of a specific vehicle can be predicted and analyzed to obtain the fault risk of the specific vehicle or provide corresponding emergency plan information support; the method and product of the present invention overcome the defect that it is difficult to make predictions when diagnosing based on single vehicle information; at the same time, the reliability or confidence of the prediction can be further improved through algorithms such as machine learning on the basis of cloud big data.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault diagnosis, and in particular relates to a driving data processing method, a monitoring device, a storage medium and a controller. Background Art

[0002] The accelerator pedal is a key component for the operator to intervene in the driving process. Its reliability is crucial to the safety and dynamic characteristics of the vehicle. Research has found that its fault prediction and diagnosis capabilities have become a bottleneck in improving the above-mentioned performance of the vehicle.

[0003] When diagnosing accelerator pedal faults, the relevant technology is usually performed on the vehicle side, which cannot provide early warning information before the fault occurs, and it is difficult to provide sufficient reference information for the operator; making it difficult to detect hidden dangers in the related operation process. Summary of the Invention

[0004] An embodiment of the present invention discloses a driving data processing method, including a first data collection and update step, a second window division and screening step, and a third feature extraction and classification step; the first data collection and update step obtains driving data of a preset number N or more target vehicles, wherein the target vehicles are all vehicles that have experienced accelerator pedal failure, and N is a sufficiently large positive integer to ensure the sample capacity; wherein the driving data includes mileage, accelerator pedal failure degree and fault code data.

[0005] Furthermore, the second window division and screening step extracts window data from the driving data of each target vehicle at a preset mileage interval D. The window data uses the current mileage of each vehicle as the right boundary and the mileage interval D as the spacing, and divides the window in the direction of decreasing mileage. Then, the third feature extraction and classification step detects the fault code data of the window data and realizes fault prediction.

[0006] Specifically, if the window after the window data is divided contains fault code data, the corresponding window is discarded and the third feature extraction and classification step is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if the fault code data exists at the right boundary of the window data, the window label corresponding to the window data is marked as a fault code window; otherwise, it is marked as a non-fault code window.

[0007] Among them, if the value or number of the fault degree data in the window divided by the window data is less than the preset fault degree threshold, then those or that windows divided by the window data are deleted to avoid individual data from interfering with the judgment or prediction process and improve the stability of the prediction.

[0008] Specifically, the fault degree can be determined by the engine management system (EMS) and / or related controller by collecting the first voltage and the second voltage at the vehicle end and filtering them; then performing a difference operation on the filtered first voltage and the second voltage; finally, dividing the intermediate result of the difference operation by the consistency diagnosis threshold of the fault degree to obtain the fault degree data; the fault code data can be obtained after diagnosis by the engine management system (EMS); wherein, the target vehicle can be a vehicle of the same model or a vehicle using the same type of engine.

[0009] A commonly used mileage interval D may be 200 kilometers, which can effectively generate sufficient samples.

[0010] Furthermore, the third feature extraction and classification step can extract feature information of the fault degree data; the feature information can be in the form of a four-feature combination, a three-feature combination, or any combination of the features in the four-feature combination or the three-feature combination; wherein the four-feature combination includes the maximum value, gradient value, mean, and standard deviation of the fault degree data; and the three-feature combination includes the upper quartile, lower quartile, and variance.

[0011] Furthermore, the driving data processing method may also include a fourth model training and optimization step, which obtains characteristic information of all or selected windows of the driving data and uses a pattern recognition algorithm and / or a machine learning process to perform model training, thereby identifying window labels; the window labels include the above-mentioned fault code windows and non-fault code windows.

[0012] Furthermore, in the application of an actual prediction system or monitoring system, the driving data processing method may also include a fifth window identification and prediction step; after the driving data is updated in the first data collection and update step, the driving data is uploaded to the cloud through the fifth window identification and prediction step, and the window is divided toward the zero mileage direction with the current data point to obtain the current value of the window data and the current value of the feature information; the model trained by the fourth model training and optimization step is used for identification, and the fault risk of the current data is judged or a preset plan response is implemented.

[0013] Corresponding to the above method, an embodiment of the present invention also discloses a monitoring device, including a first data acquisition and update unit, a second window division and screening unit, and a third feature extraction and classification unit; its first data acquisition and update unit obtains driving data of a preset number N or more target vehicles, and the target vehicles have all experienced accelerator pedal failure, N is a positive integer; wherein the driving data includes mileage, accelerator pedal failure degree and fault code data.

[0014] Furthermore, its second window division and screening unit extracts window data from the driving data of each target vehicle at a preset mileage interval D. The window data uses the current mileage of each vehicle as the right boundary and the mileage interval D as the spacing, and divides the window in the direction of decreasing mileage. Then, the third feature extraction and classification unit detects the fault code data of its window data to realize the corresponding prediction function.

[0015] Specifically, if the window after the window data is divided contains fault code data, the corresponding window is discarded and the operation of the third feature extraction and classification unit is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if there is fault code data at the right boundary of the window data, the window label corresponding to the window data is marked as a fault code window; otherwise, it is marked as a non-fault code window.

[0016] If the value or number of the fault degree data in the window divided by the window data is less than a preset fault degree threshold, the window to be deleted is deleted to improve the stability of the prediction.

[0017] Specifically, the fault degree can be determined by the engine management system EMS and / or controller by collecting the first voltage and the second voltage at the vehicle end and filtering them; then performing a difference operation on the filtered first voltage and the second voltage; finally, dividing the intermediate result of the difference operation by the consistency diagnosis threshold of the fault degree to obtain the fault degree data; and the fault code data can be obtained after diagnosis by the engine management system EMS.

[0018] Among them, the target vehicles can be vehicles of the same model or vehicles using the same type of engine; their mileage interval D can be optimized and selected according to the mileage; their third feature extraction and classification unit can extract feature information of the fault degree data; the feature information can be in the form of a four-feature combination or a three-feature combination; the four-feature combination includes the maximum value, gradient value, mean, and standard deviation of the fault degree data; and the three-feature combination includes the upper quartile, lower quartile, and variance data.

[0019] Furthermore, the monitoring device may also be provided with a fourth model training and optimization unit for obtaining characteristic information of all or selected windows of driving data and performing model training through pattern recognition algorithms and / or machine learning processes, thereby identifying window labels, which include fault code windows and non-fault code windows.

[0020] In addition, the monitoring device may also be provided with a fifth window identification and prediction unit; after its first data acquisition and update unit updates the driving data, the fifth window identification and prediction unit uploads the driving data to the cloud, and divides the window toward the zero mileage direction based on the current data point, and obtains the current value of the window data and the current value of the feature information; the model trained by the fourth model training and optimization unit is used for identification, and the fault risk of the current data is judged or a preset plan response is implemented.

[0021] An embodiment of the present invention also discloses a computer storage medium and a controller; the computer storage medium includes a storage medium body for storing a computer program; when the computer program is executed by a microprocessor, it can implement any of the above-mentioned driving data processing methods; similarly, the controller includes any of the above-mentioned monitoring devices and / or computer storage media, which can implement similar processing processes or functions.

[0022] In summary, the present invention collects historical data of multiple faulty vehicles through the cloud and constructs window samples by segmenting the historical data of each vehicle, and then predicts the fault of the target component under the drive of the window data; wherein, the target component can be an important component that sends vehicle control information, that is, the accelerator pedal; further, through feature extraction of window data and pattern recognition or machine learning of cloud vehicle data, a fault model of the target component is constructed, and the driving data updated in real time by the specific vehicle can be predicted and analyzed to obtain the fault risk of the specific vehicle or provide corresponding plan information support; the method and product of the present invention overcome the defect that it is difficult to make predictions when diagnosing based on single vehicle information; at the same time, on the basis of cloud big data, the reliability or confidence of the prediction can be further improved through algorithms such as machine learning.

[0023] It should be noted that the terms "first", "second" and similar terms used in this article are only for describing the various components of the technical solution, and do not constitute a limitation of the technical solution, nor can they be understood as an indication or suggestion of the importance of the corresponding elements; elements with terms such as "first", "second" and similar terms indicate that the corresponding technical solution contains at least one of the element. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solution of the present invention and facilitate a further understanding of the technical effects, technical features and purposes of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings. The accompanying drawings constitute an essential part of the specification and are used together with the embodiments of the present invention to illustrate the technical solution of the present invention, but do not constitute a limitation to the present invention.

[0025] The same reference numerals in the accompanying drawings represent the same components, specifically:

[0026] Figure 1The figure is a schematic diagram of the model construction process of the embodiment of the present invention.

[0027] Figure 2 Schematic diagram of the window division and screening process according to an embodiment of the present invention.

[0028] Figure 3 This is the fault degree distribution of the embodiment of the present invention.

[0029] Figure 4 Schematic diagram of window characteristic values ​​according to an embodiment of the present invention.

[0030] Figure 5 This is the dataset format of the embodiment of the present invention.

[0031] Figure 6 This is the fault degree distribution details of the embodiment of the present invention.

[0032] Figure 7 Schematic diagram of the process of the present invention.

[0033] Figure 8 It is a schematic diagram of the structure of an embodiment of the device of the present invention.

[0034] Figure 9 This is one of the schematic diagrams of the composition and layout structure of the product embodiment of the present invention.

[0035] Figure 10 This is the second schematic diagram of the composition and layout structure of the product embodiment of the present invention.

[0036] Figure 11 This is the third schematic diagram of the composition and layout structure of the product embodiment of the present invention.

[0037] in:

[0038] 100-first data collection and updating step;

[0039] 199-window data;

[0040] 200-second window division and screening step;

[0041] 210-window initialization step;

[0042] 220-window screening step;

[0043] 230-window identification step;

[0044] 291-maximum feature;

[0045] 292- gradient features;

[0046] 293-mean feature;

[0047] 294-standard deviation characteristics;

[0048] 299-window label;

[0049] 300-third feature extraction and classification step;

[0050] 399-feature information;

[0051] 400-Fourth model training and optimization step;

[0052] 500-fifth window identification and prediction step;

[0053] 600-Fault degree and fault code identification area;

[0054] 601-First fault level;

[0055] 602-Second fault level;

[0056] 603-third fault level;

[0057] 666-fault code;

[0058] 700-Fault code window example;

[0059] 701, 702, 703 - No fault code window example;

[0060] 777-The window must be deleted;

[0061] 800-Monitoring device;

[0062] 810-first data collection and update unit;

[0063] 820-second window division and screening unit;

[0064] 830-Third feature extraction and classification unit;

[0065] 840-Fourth Model Training and Optimization Unit;

[0066] 850-Fifth window recognition and prediction unit;

[0067] 900-vehicles;

[0068] 901-controller;

[0069] 903-Computer storage media. Implementation Method

[0070] The present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments. Of course, the specific embodiments described below are only for explaining the technical solutions of the present invention, rather than for limiting the present invention.

[0071] Furthermore, the parts described in the embodiments or drawings are merely examples of relevant parts of the present invention, rather than the entire present invention.

[0072] like Figure 1 、 Figure 2 、 Figure 7 The driving data processing method shown includes a first data collection and update step 100, a second window division and screening step 200, and a third feature extraction and classification step 300; wherein the first data collection and update step 100 obtains driving data of a preset number N or more target vehicles, wherein the target vehicles have all experienced accelerator pedal failure, and N is a positive integer; wherein the driving data includes mileage, accelerator pedal failure degree, and fault code data.

[0073] Furthermore, the second window division and screening step 200 extracts the driving data of each vehicle in the target vehicle at a preset mileage interval D. Figure 5 The window data 199 shown is divided into windows in the direction of decreasing mileage with the current mileage of each vehicle as the right boundary and the mileage interval D as the spacing; its third feature extraction and classification step 300 detects the fault code data of the window data 199. If the window after the window data 199 is divided contains fault code data, the corresponding window is discarded and the third feature extraction and classification step 300 is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if there is fault code data at the right boundary of the window data 199, the window label 299 corresponding to the window data 199 is marked as a fault code window; otherwise, it is marked as a non-fault code window.

[0074] If the value or number of the fault degree data in the window divided by the window data 199 is less than the preset fault degree threshold, then the window data 199 is deleted. Figure 3 The window data 199 shown is divided into a window 777 to be deleted.

[0075] Typically, the extent of the fault can be determined by the engine management system EMS and / or Figures 9 to 11 The controller 901 shown collects the first voltage and the second voltage at the vehicle end and filters them; then performs a difference operation on the filtered first voltage and the second voltage; finally, divides the intermediate result of the difference operation by the consistency diagnosis threshold of the fault degree to obtain the fault degree data; wherein, the fault code data can be obtained after diagnosis by the engine management system EMS; wherein, the target vehicle can be a vehicle 900 of the same model or a vehicle 900 using the same type of engine; specifically, its mileage interval D can be selected as 200 kilometers.

[0076] In addition, if Figure 5As shown, the third feature extraction and classification step 300 can extract feature information 399 of the fault degree data; the feature information 399 includes a four-feature combination, namely the maximum value, gradient value, mean value, and standard deviation of the fault degree data.

[0077] Furthermore, if Figure 5 、 Figure 7 As shown, the driving data processing method also includes a fourth model training and optimization step 400; the fourth model training and optimization step 400 obtains the feature information 399 of all or selected windows of the driving data and performs model training through a pattern recognition algorithm and / or a machine learning process, thereby identifying the window label 299, which includes a fault code window and a non-fault code window.

[0078] Furthermore, if Figure 7 As shown, the driving data processing method may also include a fifth window identification and prediction step 500; after the driving data is updated in the first data collection and update step 100, the fifth window identification and prediction step 500 uploads the driving data to the cloud, and divides the window toward the zero mileage direction with the current data point, and obtains the current value of the window data 199 and the current value of the feature information 399; the model trained by the fourth model training and optimization step 400 is used for identification, and the fault risk of the current data is judged or a preset plan response is implemented.

[0079] In addition, if Figure 8 The monitoring device 700 shown includes a first data acquisition and update unit 810, a second window division and screening unit 820, and a third feature extraction and classification unit 830; its first data acquisition and update unit 810 obtains driving data of a preset number N or more target vehicles, and the target vehicles have all experienced accelerator pedal failure, N is a positive integer; the driving data includes mileage, accelerator pedal failure degree and fault code data.

[0080] Its second window division and screening unit 820 extracts window data 199 with a preset mileage interval D for the driving data of each target vehicle. The window data 199 is divided into windows in the direction of decreasing mileage with the current mileage of each vehicle as the right boundary and the mileage interval D as the spacing; its third feature extraction and classification unit 830 detects the fault code data of the window data 199. If the window after the window data 199 is divided contains fault code data, the corresponding window is discarded and the operation of the third feature extraction and classification unit 830 is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if there is fault code data at the right boundary of the window data 199, the window label 299 corresponding to the window data 199 is marked as a fault code window; otherwise, it is marked as a non-fault code window.

[0081] Furthermore, if Figure 3 、 Figure 5 As shown, if the value or number of the fault degree data in the window divided by the window data 199 is less than the preset fault degree threshold, the window 777 to be deleted divided by the window data 199 is deleted.

[0082] Among them, such as Figures 9 to 11 As shown, the fault degree can be determined by the engine management system EMS and / or the controller 901 collecting the first voltage and the second voltage at the vehicle end and filtering them; then performing a difference operation on the filtered first voltage and the second voltage; finally, dividing the intermediate result of the difference operation by the consistency diagnosis threshold of the fault degree to obtain the fault degree data; the fault code data can be obtained after diagnosis by the engine management system EMS.

[0083] Specifically, the target vehicle may be a vehicle 900 of the same model or a vehicle 900 using the same type of engine; and the mileage interval may be selected as 200 kilometers.

[0084] In addition, the third feature extraction and classification unit 830 extracts feature information 399 of the fault degree data; the feature information 399 includes a four-feature combination, namely, the maximum value, gradient value, mean value, and standard deviation of the fault degree data.

[0085] Furthermore, the monitoring device 700 also includes a fourth model training and optimization unit 840; the fourth model training and optimization unit 840 can obtain the characteristic information 399 of all or selected windows of the driving data and perform model training through pattern recognition algorithms and / or machine learning processes, and then identify window labels 299, which include fault code windows and non-fault code windows.

[0086] Furthermore, if Figure 8 As shown, the monitoring device 700 also includes a fifth window identification and prediction unit 850; after its first data acquisition and update unit 810 updates the driving data, the fifth window identification and prediction unit 850 uploads the driving data to the cloud and divides the window toward the zero mileage direction based on the current data point to obtain the current value of the window data 199 and the current value of the feature information 399; and then uses the model trained by the fourth model training and optimization unit 840 for identification, thereby performing fault judgment on the fault risk of the current data or implementing a preset plan response.

[0087] like Figures 9 to 11 As shown, its computer storage medium 903 includes a storage medium body for storing computer programs; when the computer program is executed by the microprocessor, any of the above driving data processing methods can be implemented; similarly, its controller 901 includes any of the above monitoring devices 700 and / or any of the above computer storage media 903.

[0088] For Figure 3 As shown in the figure, the data from a vehicle that had an accelerator pedal failure is divided into windows of 200 kilometers. Figure 3 The distribution of mileage and fault degree shown in the figure is used to construct the embodiment. Figure 5 A sample of window data 199 is shown.

[0089] Specifically, the fault degree and detailed diagram of the fault code can be referred to Figure 6 Medium fault level and fault code identification area 600; first fault level 601, second fault level 602, third fault level 603; wherein, fault code 666 is obtained by EMS or controller.

[0090] The vehicle's current maximum mileage of 8,300 kilometers is the right boundary of the window. Windows are divided by sliding from 8,300 kilometers toward 0 kilometers. These windows are identified to see whether they contain fault code 666. At 6,200 kilometers, the window on the right boundary contains fault code 666. Therefore, the corresponding window 700 is discarded.

[0091] Then, the window is divided by sliding forward with the mileage point of the fault code 6000 kilometers as the right boundary until the mileage reaches 0; Figure 3 As shown, if the number of fault severity data in a window is less than 5, the window 777 is deleted.

[0092] Specifically, if the right border of the window is the position corresponding to fault code 666, the record window label 299 is fault code window 700; wherein, the fault code 666 at the right border of the window is deleted, and the other parts of the record window label as the no-fault-code window are retained, such as no-fault-code windows 701, 702, and 703.

[0093] Among them, such as Figure 4 As shown, four features can be extracted from the fault degree data obtained in each window, namely, maximum value 291, gradient 292, mean value 293, and standard deviation 294.

[0094] Specifically, for each vehicle that has a fault code, the effective window can be obtained by dividing the window, and the following equations can be calculated for each window: Figure 4 The four corresponding eigenvalues ​​are shown.

[0095] Furthermore, if Figure 5 The data set is constructed by using the structure shown in the figure, and the data set is trained with a machine learning model to obtain the required prediction model.

[0096] Among them, vehicle data can be uploaded to the cloud in real time, the window can be divided forward by the latest data point, and a pre-trained prediction model can be used to identify whether the window is a fault code window, thereby achieving early prediction of accelerator pedal failure.

[0097] Compared with the related art, the embodiments of the present invention only perform fault diagnosis of a single vehicle accelerator pedal on the vehicle side. The method and product of the present invention, on the one hand, jointly analyze the accelerator pedal data of multiple vehicles on the cloud, and can perform targeted statistical analysis on the accelerator pedal of a vehicle model or an engine type, so as to better analyze the operating status of the accelerator pedal; on the other hand, it can realize the early prediction of accelerator pedal failure, which can effectively help car owners maintain the accelerator pedal while reducing driving safety issues introduced by accelerator pedal failure, thereby improving the driving experience of the operator.

[0098] The content of feature extraction is not limited to statistics of the maximum value 291, gradient 292, mean 293, and standard deviation 294. The upper quartile, lower quartile, and variance may also be used to replace the above four eigenvalues.

[0099] It should be noted that the above embodiments are only for the purpose of more clearly illustrating the technical solutions of the present invention. Those skilled in the art will understand that the implementation methods of the present invention are not limited to the above contents, and obvious changes, replacements or substitutions based on the above contents do not exceed the scope covered by the technical solutions of the present invention; other implementation methods will also fall within the scope of the present invention without departing from the concept of the present invention.

Claims

1. A driving data processing method, characterized in that: The invention comprises a first data collection and updating step (100), a second window division and screening step (200), and a third feature extraction and classification step (300); wherein the first data collection and updating step (100) obtains driving data of a preset number N or more target vehicles, wherein the target vehicles have all experienced accelerator pedal failure, and N is a positive integer; the driving data includes mileage, accelerator pedal failure degree, and fault code data; the second window division and screening step (200) extracts window data (199) at a preset mileage interval D for the driving data of each vehicle in the target vehicles, and the window data (199) is divided into two parts based on the current mileage of each vehicle. The mileage is the right boundary, and the windows are divided in the direction of decreasing mileage with the mileage interval D as the spacing; the third feature extraction and classification step (300) detects the fault code data of the window data (199), and if the window after the window data (199) is divided contains the fault code data, the corresponding window is discarded and the third feature extraction and classification step (300) is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if the fault code data exists at the right boundary of the window data (199), the window label (299) corresponding to the window data (199) is marked as a fault code window; otherwise, it is marked as a non-fault code window.

2. The driving data processing method according to claim 1, wherein: If the value or number of the fault degree data in the window divided by the window data (199) is less than a preset fault degree threshold, the window to be deleted (777) divided by the window data (199) is deleted.

3. The driving data processing method according to claim 2, wherein: The fault degree is first collected by the engine management system EMS and / or the controller (901) and the first and second voltages of the vehicle end are filtered; Then, performing a difference operation on the filtered first voltage and the second voltage; finally, dividing an intermediate result of the difference operation by a consistency diagnosis threshold of the fault degree to obtain fault degree data; The fault code data is obtained after diagnosis by the engine management system EMS.

4. The driving data processing method according to any one of claims 1 to 3, wherein: The target vehicles are vehicles (900) of the same model or vehicles (900) using the same type of engine.

5. The driving data processing method according to claim 4, wherein: The mileage interval D=200, and the dimension of the mileage interval D is kilometers, miles and / or nautical miles.

6. The driving data processing method according to any one of claims 1 to 3 or 5, wherein: The third feature extraction and classification step (300) extracts feature information (399) of the fault degree data; the feature information (399) includes a four-feature combination, a three-feature combination and / or a combination of the four-feature combination and each feature in the three-feature combination; the four-feature combination includes the maximum value, gradient value, mean, and standard deviation of the fault degree data; the three-feature combination includes the upper quartile, lower quartile, and variance.

7. The driving data processing method as described in claim 6 further includes a fourth model training and optimization step (400); the fourth model training and optimization step (400) obtains the feature information (399) of all or selected windows of the driving data and performs model training through a pattern recognition algorithm and / or a machine learning process, thereby identifying the window label (299), and the window label (299) includes the fault code window and the non-fault code window.

8. The driving data processing method according to claim 7 further includes a fifth window identification and prediction step (500); after the first data acquisition and update step (100) updates the driving data, the fifth window identification and prediction step (500) uploads the driving data to the cloud, and divides the window toward the zero mileage direction based on the current data point, obtains the current value of the window data (199) and the current value of the feature information (399); uses the model trained by the fourth model training and optimization step (400) for identification, and performs fault judgment on the fault risk of the current data or implements a preset plan response.

9. A monitoring device (700), comprising a first data acquisition and updating unit (810), a second window division and screening unit (820), and a third feature extraction and classification unit (830); wherein: The first data acquisition and updating unit (810) acquires driving data of a preset number N or more target vehicles, wherein the target vehicles all experience accelerator pedal failure, where N is a positive integer; the driving data includes mileage, accelerator pedal failure degree, and fault code data; The second window division and screening unit (820) extracts window data (199) at a preset mileage interval D for the driving data of each vehicle in the target vehicle, and the window data (199) is divided into windows in a direction of decreasing mileage with the current mileage of each vehicle as the right boundary and the mileage interval D as the spacing; the third feature extraction and classification unit (830) detects the fault code data of the window data (199); if the window after the window data (199) is divided contains the fault code data, the corresponding window is discarded and the operation of the third feature extraction and classification unit (830) is repeated with the mileage point corresponding to the fault code data as the right boundary until the mileage is 0; if the fault code data exists at the right boundary of the window data (199), the window label (299) corresponding to the window data (199) is marked as a fault code window; otherwise, it is marked as a non-fault code window.

10. The monitoring device (700) according to claim 9, wherein: If the value or number of the fault degree data in the window divided by the window data (199) is less than a preset fault degree threshold, the window to be deleted (777) divided by the window data (199) is deleted.

11. The monitoring device (700) according to claim 10, wherein: The fault degree is first collected by the engine management system EMS and / or the controller (901) and the first and second voltages of the vehicle end are filtered; Then, performing a difference operation on the filtered first voltage and the second voltage; finally, dividing an intermediate result of the difference operation by a consistency diagnosis threshold of the fault degree to obtain fault degree data; The fault code data is obtained after diagnosis by the engine management system EMS.

12. The monitoring device (700) according to any one of claims 9 to 11, wherein: The target vehicles are vehicles (900) of the same model or vehicles (900) using the same type of engine; the mileage interval D=200, and the dimension of the mileage interval D is kilometers, miles and / or nautical miles.

13. The monitoring device (700) according to claim 12, wherein: The third feature extraction and classification unit (830) extracts feature information (399) of the fault degree data; the feature information (399) includes a four-feature combination, a three-feature combination and / or a combination of the four-feature combination and each feature in the three-feature combination; the four-feature combination includes the maximum value, gradient value, mean, and standard deviation of the fault degree data; the three-feature combination includes the upper quartile, lower quartile, and variance.

14. The monitoring device (700) as described in claim 13 further includes a fourth model training and optimization unit (840); the fourth model training and optimization unit (840) obtains the feature information (399) of all or selected windows of the driving data and performs model training through a pattern recognition algorithm and / or a machine learning process, thereby identifying the window label (299), and the window label (299) includes the fault code window and the non-fault code window.

15. The monitoring device (700) according to claim 14 further comprises a fifth window identification and prediction unit (850); after the first data acquisition and update unit (810) updates the driving data, the fifth window identification and prediction unit (850) uploads the driving data to the cloud, and divides the window toward the zero mileage direction based on the current data point, obtains the current value of the window data (199) and the current value of the characteristic information (399); uses the model trained by the fourth model training and optimization unit (840) for identification, and performs fault judgment on the fault risk of the current data or implements a preset plan response.

16. A computer storage medium (903), comprising a storage medium body for storing a computer program; when the computer program is executed by a microprocessor, the driving data processing method according to any one of claims 1 to 8 is implemented.

17. A controller (901) comprising the monitoring device (700) according to any one of claims 9 to 15 and / or the computer storage medium (903) according to claim 16.

Images