State diagnosis apparatus and method of moving system part

Patent metadata

Field	Value
Publication number	US-12154392-B2
Application number	US-202318221786-A
Country	US
Kind code	B2
Filing date	Jul 13, 2023
Priority date	Jan 6, 2020
Publication date	Nov 26, 2024
Grant date	Nov 26, 2024

Abstract

A state diagnosis apparatus of a moving system part is provided and includes a sensor unit that measures and collects state data of a moving system part relevant to an engine. A graphic controller primarily diagnoses classification data generated by classifying the state data according to a predetermined filtering condition as a normal state or an abnormal state using a Deep Learning model.

First claim

What is claimed is:

1. A state diagnosis apparatus of a moving system part, comprising:
a sensor unit configured to measure and collect state data of a moving system part relevant to an engine; and
a graphic controller configured to:
primarily diagnose classification data generated by classifying the state data based on a predetermined filtering condition as a normal state or an abnormal state using a Deep Learning model, and
perform secondary diagnosis to calculate a normal level grade in response to diagnosing the normal state.

2. The state diagnosis apparatus of the moving system part of claim 1 , wherein the state data includes noise data and vibration data of the moving system part.

3. The state diagnosis apparatus of the moving system part of claim 2 , wherein the state data includes noise data and vibration data of the abnormal state generated by a failure of the moving system part, and includes noise data and vibration data by mileage according to the normal state of the engine.

4. The state diagnosis apparatus of the moving system part of claim 3 , wherein the state data is collected when the engine is an idle condition and a predetermined constant speed.

5. The state diagnosis apparatus of the moving system part of claim 1 , wherein the Deep Learning model is a structure including Recurrent Neural Network (RNN), Attention Mechanism, and Deep Neural Network (DNN).

6. The state diagnosis apparatus of the moving system part of claim 1 , wherein the normal level grade is a value based on a Noise, Vibration, Harshness (NVH) level of the engine.

7. The state diagnosis apparatus of the moving system part of claim 6 , wherein the normal level grade is expressed by an energy relative ratio in a frequency band in which percussive among the state data is generated larger than a predetermined reference value, an energy relative ratio in a frequency band which governs sound pressure of low frequency to high frequency bands, and a score summed by multiplying an extracted result value extracted by classifying a percussive sound component in the corresponding frequency band by each weight.

8. The state diagnosis apparatus of the moving system part of claim 7 , wherein the extracted result value is calculated using a Harmonic-Percussive Source Separation (HPSS) algorithm.

9. The state diagnosis apparatus of the moving system part of claim 1 , wherein the graphic controller is configured to execute a variable cylinder management (VCM) evaluation mode which adds an acceleration condition for re-measuring the state information.

10. The state diagnosis apparatus of the moving system part of claim 1 , wherein the engine is an internal combustion engine, an electric motor, a gas turbine, or an electric generator.

11. A state diagnosis method of a moving system part, comprising:
collecting, by a sensor unit, state data of a moving system part relevant to an engine;
classifying, by a graphic controller, the state data based on a predetermined filtering condition to generate classification data;
primarily diagnosing, by the graphic controller, the classification data as a normal state or an abnormal state using a Deep Learning model, and
performing, by the graphic controller, secondary diagnosis to calculate a normal level grade in response to diagnosing the normal state.

12. The state diagnosis method of the moving system part of claim 11 , wherein the state data includes noise data and vibration data of the moving system part.

13. The state diagnosis method of the moving system part of claim 12 , wherein the state data includes noise data and vibration data of the abnormal state generated by a failure of the moving system part, and includes noise data and vibration data by mileage according to the normal state of the engine.

14. The state diagnosis method of the moving system part of claim 13 , wherein the state data is collected when the engine is an idle condition and a predetermined constant speed.

15. The state diagnosis method of the moving system part of claim 11 , wherein the Deep Learning model is a structure including Recurrent Neural Network (RNN), Attention Mechanism, and Deep Neural Network (DNN).

16. The state diagnosis method of the moving system part of claim 11 , wherein the normal level grade is a value based on a Noise, Vibration, Harshness (NVH) level of the engine.

17. The state diagnosis method of the moving system part of claim 16 , wherein the normal level grade is expressed by an energy relative ratio in a frequency band in which percussive sound is generated larger than a predetermined reference value among the state data, an energy relative ratio in a frequency band which governs sound pressure of low frequency to high frequency bands, and a score summed by multiplying an extracted result value extracted by classifying the percussive sound component in the corresponding frequency band by each weight.

18. The state diagnosis method of the moving system part of claim 11 , wherein the engine is an internal combustion engine, an electric motor, a gas turbine, or an electric generator.

Assignees

• Hyundai Motor Co Ltd
• Kia Corp

Inventors

• Lee Dong-Chul
• Jung In-Soo

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title