Method of AI-based vehicle diagnosis using CAN data and device thereof

Patent metadata

Field	Value
Publication number	US-11721141-B2
Application number	US-202117376973-A
Country	US
Kind code	B2
Filing date	Jul 15, 2021
Priority date	Feb 19, 2021
Publication date	Aug 8, 2023
Grant date	Aug 8, 2023

Abstract

A device for AI-based vehicle diagnosis using CAN data may include an engine; a vibration sensor mounted in an engine compartment in which the engine is mounted and configured for detecting a vibration signal; and a controller area network (CAN) communicating with one or more of an environmental condition, a vehicle status, an engine status, and an engine control parameter, wherein data preprocessing from the vibration sensor and the CAN is performed to determine features in which correlation between vibration data (dB) exceeding a threshold value of irregular vibrations being generated by the engine and the CAN data is equal to or greater than 90%.

First claim

What is claimed is:

1. A device for AI-based vehicle diagnosis using CAN data, the device comprising:
an engine;
a vibration sensor mounted in an engine compartment in which the engine is mounted and configured for detecting a vibration signal; and
a controller area network (CAN) communicating with one or more of an environmental condition, a vehicle status, an engine status, and an engine control parameter,
wherein data preprocessing from the vibration sensor and the CAN is performed to determine features in which correlation between vibration data (dB) exceeding a threshold value of irregular vibrations being generated by the engine and the CAN data is equal to or greater than 90%,
wherein the threshold value of the irregular vibrations is subject to 0.5th and 1st order components.

2. The device according to claim 1 , wherein the preprocessing of the vibration data includes distinguishing first vibration data having a level of the vibration signal which is equal to or greater than a reference dB and continuing for a first time duration or more than the first time duration.

3. The device according to claim 2 , wherein the preprocessing of the vibration data further includes additionally distinguishing second vibration data having an average level of the vibration signal which is equal to or greater than the reference dB and continuing the average level for a second time duration or more than the second time duration and for a third time duration or less than the third time duration.

4. The device according to claim 3 , wherein the preprocessing of the vibration data further includes removing third vibration data having the average level of the vibration signal which is lower than the reference dB and continuing for less than the third time duration.

5. The device according to claim 1 , wherein the preprocessing of the vibration data includes extending by one frame left and right from a frame in which a label for an upper predetermined percentage of the vibration level is located after obtaining vibration levels in the 0.5th and 1st orders of the engine and a number distribution by the vibration levels.

6. The device according to claim 5 , wherein the preprocessing of the vibration data further includes integrating extended labels in the 0.5th and 1st orders, and integrating frame data of a lower predetermined time into one label from a count (y axis) for a number of frames (x axis) corresponding to a distance between the irregular vibrations.

7. The device according to claim 5 , wherein with respect to the label extended by one frame left and right, a slope value having an inclination left and right is provided to the label by applying average filtering for three labels.

8. The device according to claim 1 , wherein the preprocessing of the CAN data includes removing a target label factor having a low influence using a variance value and an average value among the features of the CAN data.

9. A method for AI-based vehicle diagnosis using CAN data, the method comprising:
obtaining the CAN data from a controller area network (CAN) communicating with one or more of an environmental condition, a vehicle status, an engine status, and an engine control parameter;
obtaining vibration data by detecting, by a vibration sensor, a vibration signal from an engine generating idle vibrations in a vehicle; and
performing data preprocessing for obtaining, from the vibration data or the CAN data, features in which correlation between the vibration data (dB) exceeding a threshold value of irregular vibrations among 0.5th and 1st order levels of the engine and the CAN data is equal to or greater than 90%,
wherein the threshold value of the irregular vibrations is subject to 0.5th and 1st order components.

10. The method according to claim 9 , wherein the preprocessing of the vibration data includes additionally distinguishing:
first vibration data having a level of the vibration signal which is equal to or greater than a reference dB and continuing for a first time duration or more than the first time duration,
second vibration data having an average level of the vibration signal which is equal to or greater than the reference dB and continuing the average level for a second time duration or more than the second time duration and for a third time duration or less the third time duration, and
a third vibration data having the average level of the vibration signal which is lower than the reference dB and continuing for less than the third time duration,
wherein the preprocessing of the vibration data further includes removing the third vibration data.

11. The method according to claim 9 , wherein the preprocessing of the CAN data includes removing a target label factor having a low influence in case that a variance value is 0 with respect to a target label using the variance value and an average value among all the collected CAN data features.

12. The method according to claim 11 , wherein the preprocessing of the CAN data further includes removing features in which a minimum value and a maximum value of the CAN data are equal to each other by applying a Jenson-Shannon divergence (JSD) method.

13. The method according to claim 12 , wherein the preprocessing of the CAN data further includes selecting the features of the CAN data by determining a Pearson coefficient between the CAN data features and a target while reducing the features using physical features of the CAN data in a Pearson correlation analysis method.

14. A device for AI-based vehicle diagnosis and control using CAN data, the device comprising:
an engine;
a vibration sensor mounted in an engine compartment in which the engine is mounted and configured for detecting a vibration signal;
a controller area network (CAN) communicating with one or more of an environmental condition, a vehicle status, an engine status, and an engine control parameter; and
a diagnosis and control unit;
wherein data preprocessing from the vibration sensor and the CAN is performed to determine features in which correlation between vibration data (dB) exceeding a threshold value of irregular vibrations being generated by the engine and the CAN data is equal to or greater than 90%,
wherein when the CAN data obtained from another vehicle of a same kind as the vehicle is inputted to the diagnosis and control unit, the diagnosis and control unit changes a control factor to remove output AI-based irregular vibrations with respect to a diagnosis result of the irregular vibrations, and
wherein the threshold value of the irregular vibrations is subject to 0.5th and 1st order components.

Assignees

• Hyundai Motor Co Ltd
• Kia Corp
• Iucf Hyu

Inventors

• Lee Dong-Chul
• Jung In-Soo
• Jeon Dong-Yeoup
• Chang Joon-Hyuk

Classifications

• B60Y2306/15

  Failure diagnostics · CPC title

• G06N20/20

  Ensemble learning · CPC title

• B60W50/0205 Primary

  Diagnosing or detecting failures; Failure detection models · CPC title

• G07C5/0808 Primary

  Diagnosing performance data (testing of vehicles G01M17/00; testing of electrical installation on vehicles G01R31/005) · CPC title

• G01M15/12 Primary

  by monitoring vibrations · CPC title