Injector abnormal cylinder diagnosis method and injector abnormality diagnosis system based on signal deviation

Patent metadata

Field	Value
Publication number	US-11236694-B2
Application number	US-202017110506-A
Country	US
Kind code	B2
Filing date	Dec 3, 2020
Priority date	Jun 8, 2020
Publication date	Feb 1, 2022
Grant date	Feb 1, 2022

Abstract

An injector failure cylinder diagnosis method based on signal deviation of an injector failure diagnosis system measures, by a controller, noise/vibration signals due to the combustion of an engine, separates an injector abnormal frequency band signal as an injector signal from the noise/vibration signals, divides cylinder number time series data, which use a signal maximum value of the noise/vibration signals as a cylinder #1, into segments, extracts a feature vector after confirming a segment number of the cylinder #1 with a vibration reduction signal of the noise/vibration signals re-measured using one of the cylinders as an idle cylinder and sorting it by injector causing vibration time series data, and confirms a failure injector with the feature vector, thereby independently diagnosing whether the injector for each cylinder is normal or abnormal considering the deviation of the noise/vibration signals between the injectors for each cylinder from the measured noise and vibration signals.

First claim

The invention claimed is:

1. An injector abnormal cylinder diagnosis method, the method comprising:
separating, by a controller, an injector signal with respect to an injector abnormal frequency band signal indicating abnormality of an injector from noise or vibration signals due to the combustion of an engine;
dividing cylinder number time series data, which uses a signal maximum value of the noise or vibration signals as a cylinder #1, into segments, and sorting injector causing vibration time series data by confirming a segment number using, as the cylinder #1, a vibration reduction signal of the noise/vibration signals measured in one idle cylinder state;
extracting a feature vector with respect to each of the segments sorted by the injector causing vibration time series data; and
confirming an abnormal injector by processing the feature vector based on artificial intelligence (AI), or confirming the abnormal injector by processing the feature vector based on a threshold.

2. The method of claim 1 , wherein the noise or vibration signals are measured by an accelerometer.

3. The method of claim 1 , wherein the separating of the injector signal comprises:
measuring the noise or vibration signals;
confirming the injector abnormal frequency band signal using the noise or vibration signals as raw data; and
separating an injector signal indicating the abnormality of the injector by a filter processing.

4. The method of claim 1 , wherein the sorting of the injector causing vibration time series data comprises:
confirming the signal maximum value as a maximum value among signal magnitude values of the noise or vibration signals;
confirming the number of cylinders of the engine;
dividing the cylinder number time series data into the segments using, as a cylinder #1, a cylinder in which the signal maximum value is generated, and sorting the segments; and
confirming an idle cylinder-based number which confirms the segment number by assigning the idle cylinder in which the combustion causing vibration is reduced to the cylinder #1; or
confirming a sensor-based number which confirms the segment number by assigning the idle cylinder to the cylinder #1 in which a minimum signal of a crankshaft signal or a camshaft signal is generated.

5. The method of claim 4 , wherein the sorting of the segment comprises:
setting a segment #1 by assigning a length of the segment to the signal maximum value;
setting a second segment according to the number of cylinders by increasing the length of the segment; and
assigning a cylinder number by setting the segment #1 and the second segment.

6. The method of claim 5 , wherein the setting of the segment #1 is applied with ½ of the length of the segment.

7. The method of claim 4 , wherein the confirming of the idle cylinder-based number comprises:
setting the idle cylinder as the cylinder #1;
deactivating the injector #1 of the idle cylinder;
re-measuring the noise or vibration signals of the engine;
sorting the re-measured noise or vibration signals with injector causing time series data;
confirming a segment with a vibration minimum value among the injector causing time series data as the segment of the idle cylinder due to the reduction in the combustion causing vibration; and
numbering the segment using, as the cylinder #1, a cylinder in which a combustion causing vibration acceleration minimum value is generated.

8. The method of claim 4 , wherein the confirming of the sensor-based cylinder number comprises:
confirming a signal strength magnitude for the crankshaft signal or the camshaft signal; and
confirming the segment number using, as the cylinder #1, a cylinder in which a minimum signal strength applied as a reference point among the signal strength magnitudes is generated.

9. The method of claim 1 , wherein the extracting of the feature vector comprises:
dividing the segments into segment units;
setting the number of feature vectors for the segment units; and
applying an extracted feature vector as a feature vector for abnormality diagnosis by extracting the number of feature vectors.

10. The method of claim 9 , wherein the number of feature vectors is evenly applied to a group of the segment units.

11. The method of claim 10 , wherein the number of feature vectors is n×100 or more (n is an integer of 1 or more) which may be diagnosed.

12. The method of claim 9 , wherein the extracting of the feature vector is performed by any one of a vibration maximum value, a root mean square (RMS), a variance, a peak-to-peak, a kurtosis, and a crest factor.

13. The method of claim 1 , wherein the confirming of the AI-based abnormal injector comprises:
constructing a Deep Neural Network (DNN) model by extracting the feature vector as the feature vector for the abnormality diagnosis, and using the feature vector for the abnormality diagnosis as an input in an AI system;
performing the AI using Machine learning or Deep learning with respect to the DNN model;
determining, by the controller, the improvement in the accuracy of the DNN model using the output of the DNN model as the result of the DNN model;
confirming the abnormal injector from the result of the DNN model; and
outputting the abnormal injector confirmation as a diagnosis result.

14. The method of claim 13 , wherein the confirming of the AI-based abnormal injector comprises:
re-constructing the Deep Neural Network (DNN) model by increasing the number of feature vectors for the abnormality diagnosis extracted in response to the demand for the improvement in the accuracy of the DNN model.

15. The method of claim 1 , wherein the confirming of the threshold-based abnormal injector comprises:
performing, by a data mapper, cylinder mapping by sorting the time series data of the feature vector for each cylinder;
matching the threshold for each cylinder with the cylinder mapping result;
confirming, by the controller, an abnormal injector from the threshold matching result; and
outputting the abnormal injector confirmation as the diagnosis result.

16. An injector abnormality diagnosis system comprising:
a controller configured to measure noise or vibration signals due to the combustion of an engine, separate an injector abnormal frequency band signal as an injector signal from the noise/vibration signals, divide cylinder number time series data, which use a signal maximum value of the noise or vibration signals as a cylinder #1, into segments, extract a feature vector after confirming a segment number of the cylinder #1 with a vibration reduction signal of the noise or vibration signals re-measured using one of the cylinders as an idle cylinder, and sort it by injector causing vibration time series data, and confirms an abnormal injector with the feature vector; and
an accelerometer configured to measure the noise or vibration signals.

17. The injector abnormality diagnosis system of claim 16 , wherein the accelerometer is attached to a location of a cylinder in which a highest signal-to-noise (S/N) ratio is measured in the engine.

18. The injector abnormality diagnosis system of claim 16 , wherein the controller is connected to an artificial intelligence (AI) system, and wherein the artificial intelligence (AI) system transfers an abnormal injector processing result of processing the feature vector based on artificial intelligence (AI) to the controller.

19. The injector abnormality diagnosis system of claim 16 , wherein the controller is connected to a data mapper, and wherein the data mapper transfers an abnormal

Assignees

• Hyundai Motor Co Ltd
• Kia Motors Corp

Inventors

• Lee Seung-Hyun
• Jung In-Soo
• Lee Dong-Chul

Classifications

• Y02T10/40

  Engine management systems · CPC title

• F02D2200/025

  Engine noise, e.g. determined by using an acoustic sensor · CPC title

• F02D2041/224

  Diagnosis of the fuel system · CPC title

• F02D41/221 Primary

  relating to the failure of actuators or electrically driven elements · CPC title

• F02D2041/1432

  the system including a filter, e.g. a low pass or high pass filter · CPC title