Method and device for diagnosing problematic noise source based on big data information

Abstract

A method for diagnosing a problematic noise source based on big data information include: measuring noise data of a powertrain of a vehicle by using a real-time noise measurement device, and converting the noise data into a signal that can be input to a portable device for diagnosing the problematic noise source through an interface device; analyzing a noise through a deep learning algorithm of an artificial intelligence on a converted signal, and diagnosing the problematic noise source as a cause of the noise; and displaying the cause of the noise by outputting a diagnostic result as the problematic noise source, and transmitting the diagnostic result to the portable device.

Claims

1. A method for diagnosing a problematic noise source based on big data information, comprising: measuring noise data of a powertrain of a vehicle by using a real-time noise measurement device and vibration data of the vehicle using a vibration sensor, and acquiring an engine speed (RPM) from controller area network (CAN) data of the vehicle, wherein the noise data, the vibration data or the engine speed of the vehicle is stored at external storage data outside the vehicle; converting the noise data, the vibration data, or the engine speed into a signal that can be input to a portable device for diagnosing the problematic noise source through an interface device, wherein the noise data, the vibration data, or the engine speed of the vehicle is resampled to obtain the converted signals, and a frequency of resampling is twice a maximum frequency of a problematic noise; analyzing the noise through a deep learning algorithm of an artificial intelligence on converted signals of the noise data, the vibration data, and the engine speed, and diagnosing, from the converted signals, the problematic noise source as a cause of the noise; and displaying the cause of the noise by outputting a diagnostic result as the problematic noise source, and transmitting the diagnostic result to the portable device.

2. The method according to claim 1, wherein the noise data is acquired through a microphone and the vibration data is acquired through the vibration sensor.

3. The method according to claim 2, wherein a bidirectional method and a gated recurrent unit (GRU) technique are applied to the deep learning algorithm.

4. The method according to claim 3, wherein an attention mechanism technique is applied to the deep learning algorithm.

5. The method according to claim 3, wherein an early stage ensemble learning technique is applied to the deep learning algorithm.

6. The method according to claim 5, wherein the artificial intelligence is configured to reproduce a sound for identification of the problematic noise source.

7. The method according to claim 1, wherein a bidirectional method and a gated recurrent unit (GRU) technique are applied to the deep learning algorithm.

8. The method of claim 7, wherein an attention mechanism technique is applied to the deep learning algorithm.

9. The method according to claim 8, wherein an early stage ensemble learning technique is applied to the deep learning algorithm.

10. The method according to claim 9, wherein the artificial intelligence is configured to reproduce a sound for identification of the problematic noise source.

11. The method according to claim 1, wherein the external storage data outside the vehicle is transmitted to the controller area network (CAN) data of the vehicle by Bluetooth.

12. A device for diagnosing a problematic noise source based on big data information, comprising: a microphone for measuring noise data of a vehicle; a vibration sensor for acquiring vibration data of the vehicle; a controller area network (CAN) module for acquiring an engine speed of the vehicle; and a controller for converting the noise data, the vibration data, and data regarding the engine speed of the vehicle into codes and receiving a diagnosis result from an artificial intelligence, wherein the noise data or the vibration data of the vehicle is stored at external storage data outside the vehicle, wherein the noise data, the vibration data, or the engine speed of the vehicle is resampled to obtain the converted codes, wherein a frequency of resampling is twice a maximum frequency of a problematic noise, and wherein the artificial intelligence diagnoses the problematic noise source based on the converted codes of the noise data, the vibration data or the engine speed by deep learning.

13. A device for diagnosing a problematic noise source based on big data information, comprising: an input data collector for collecting input data, the input data collector comprising: a microphone for measuring noise data of a vehicle; a vibration sensor for acquiring vibration data of the vehicle; a controller area network (CAN) module for acquiring an engine speed of the vehicle; and a memory for storing problematic noise regions from the noise data, the vibration data, and data regarding engine speed, wherein the noise data or the vibration data of the vehicle is stored at external storage data outside the vehicle; a controller for converting the noise data or vibration data into codes and receiving a diagnosis result from an artificial intelligence, wherein the noise data, the vibration data, or the engine speed of the vehicle is resampled to obtain the converted codes, and a frequency of resampling is twice a maximum frequency of a problematic noise; and a problematic noise diagnosis unit for diagnosing the problematic noise source based on the codes by deep learning of the artificial intelligence, wherein the artificial intelligence diagnoses the problematic noise source based on converted codes of the noise data, the vibration data or the engine speed by the deep learning, wherein the input data collector and the problematic noise diagnosis unit are separable from each other.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The above and other objects, features and other advantages of the present disclosure will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, in which:

(2) FIG. 1 is a flowchart illustrating general flow of a method for diagnosing a problematic noise source based on big data information according to an exemplary embodiment of the present disclosure;

(3) FIG. 2 is an algorithm diagram according to an exemplary embodiment of the present disclosure; and

(4) FIG. 3 is a diagram illustrating another exemplary embodiment of the present disclosure, in which an input data collector and a problematic noise diagnosis unit can be separated from each other.

DESCRIPTION OF SPECIFIC EMBODIMENTS

(5) The present disclosure may be implemented in many altered forms and have various embodiments. Therefore, an exemplary specific embodiment is illustrated in the drawings and described in detail in the following description. However, the disclosure herein is not intended to limit the present disclosure to a specific embodiment and should be understood as covering all modifications, equivalents and substitutions falling within the spirit and scope of the disclosure.

(6) Like reference numerals are used to designate like elements throughout the drawings and the description with reference to the drawings.

(7) Although the terms “first,” “second,” etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used merely to distinguish one element from another.

(8) For example, a first component may be designated as a second component and similarly a second component may be designated as a first component without departing from the scope of the present disclosure. The term “and/or” includes all combinations or any of a plurality of the associated listed items.

(9) All terms including technical or scientific terms, unless being defined otherwise, have the same meaning generally understood by a person having ordinary skill in the art pertaining to the present disclosure.

(10) The terms that are the same as the ones defined in a commonly-used dictionary should be interpreted as including the meaning consistent with the meaning in the context of the related art and should not be interpreted as being ideally or excessively formal meaning unless they are explicitly defined otherwise herein.

(11) Hereinafter, a method for diagnosing a problematic noise source based on big data information according to an exemplary embodiment of the present disclosure is described.

(12) First, referring to FIG. 1, flow of the method for diagnosing a problematic noise source based on big data information according to the present disclosure is described.

(13) FIG. 1 is a flowchart illustrating general flow of a method for diagnosing a problematic noise source based on big data information according to an exemplary embodiment of the present disclosure.

(14) FIG. 2 is an algorithm diagram of a device for diagnosing a problematic noise source according to an exemplary embodiment of the present disclosure.

(15) Noise and vibration data of a vehicle are firstly input by an input data collector 100 in step S1.

(16) The noise or vibration data of the vehicle is converted into noise and then the noise is converted into a first code by a controller of a portable device in step S2.

(17) Then, information of engine speed emerged from CAN data of the vehicle is converted into the first code by the controller in step S3.

(18) Afterwards, a problematic noise source out of the noise or vibration is diagnosed based on the first code by deep learning of an artificial intelligence in step S4.

(19) An exemplary embodiment of the present disclosure is described firstly.

(20) FIG. 2 is a control logic according to an exemplary embodiment of the present disclosure. A device for diagnosing a problematic noise source of a vehicle 1 for a control logic according to an exemplary embodiment includes an input data collector 100, an external storage data 200, a portable device 300 (for example, a smart phone and a notebook), and an artificial intelligence 420.

(21) The input data collector 100 comprises a vibration sensor 111, a microphone 112 and a Controller Area Network (CAN) module 113. The external storage data 200 comprises a memory 201 and cloud network 202.

(22) The portable device 300 for diagnosing problematic noise includes a controller 300-1 and a problematic noise diagnosis unit 310. The controller 300-1 comprises an Integrated Circuit Piezoelectric (ICP) amplifier 321, an Analog/Digital (AD) board 322, a resampling unit 323, a CAN processing unit 324, a Revolutions Per Minute (RPM) information unit 325, a noise converter 326, a first ASCII conversion units 327, a second ASCII conversion units 328, a display unit 421 and a sound reproducing unit 422. The problematic noise diagnosis unit 310 may be a processor such as a computer, a central processing unit (CPU), etc.

(23) The artificial intelligence 420 performs deep learning. Deep learning includes Bidirectional method, Gated Recurrent Unit, Recurrent Neural Network, Attention Mechanism, Bidirectional RNN (GRU), and Early stage ensemble, each of which is known in the art and therefore a detailed description thereof is omitted herein.

(24) An algorithm according to an exemplary embodiment of the present disclosure generally follows the following procedures.

(25) The noise data can be acquired through a microphone 112 in real-time while the vibration data can be acquired through a vibration sensor 111.

(26) The vibration sensor 111 may be an accelerometer sensor.

(27) One or more microphones 112 may be provided.

(28) The ICP amplifier 321 is connected to the vibration sensor 111 and the microphone 112. The ICP amplifier 321 amplifies the vibration data input from the vibration sensor 111 and the noise data input from the microphone 112, and transmits the amplified noise and vibration data to the AD board 322.

(29) The AD board 322 is an AD converter that converts the noise data and vibration data, which are analog signal, into signal of a digital form enabling learning of an artificial intelligence.

(30) The noise converter 326 is connected to the AD board 322.

(31) The CAN module 113 is connected to the CAN processing unit 324 of the controller 300-1. The CAN module 113 is a module for measuring operation conditions of the vehicle and a power train, which transmits a digital signal.

(32) These noise data, vibration data, and CAN data are all converted into a first signal form in the first and second ASCII conversion units 327 and 328.

(33) In other words, the CAN processing unit 324 outputs RPM information from the data input from the CAN module 113 to the ASCII conversion unit 327.

(34) On the other hand, the noise data and vibration data are transmitted to the second ASCII conversion unit 328 via the noise converter 326.

(35) At this time, the resampling unit 323 samples twice the frequency input from the external storage data 200 and transmits the sampling frequency to the first ASCII conversion unit 327.

(36) The first signal converted in the first and second ASCII conversion units 327 and 328 is input to the artificial intelligence 420 and then the artificial intelligence 420 diagnoses a problematic noise source out of the noise or vibration.

(37) The diagnosis result may be provided visually through a display unit 421 and a result resulting from application of the problematic noise to a band pass filter (BPF) in a sound reproducing unit 422 may be provided audibly.

(38) On the other hand, the artificial intelligence may output the result of diagnosing the problematic noise source through the problematic noise diagnosis unit 310 of the portable device 300 and is capable of reproducing a sound for identification.

(39) The portable device 300 refers to the controller 300-1 and the problematic noise diagnosis units 310.

(40) Hereinafter, another exemplary embodiment of the present disclosure is described.

(41) FIG. 3 is a control logic according to another exemplary embodiment of the present disclosure, in which an input data collector 100 and a problematic noise diagnosis unit 320 can be separated from each other.

(42) A device for diagnosing a problematic noise source of a vehicle 1 for a control logic according to an exemplary embodiment includes an input data collector 100, an external storage data 200, a portable device 300 and an artificial intelligence 420. In this case, the portable device 300 is capable of driving Window 10, an artificial intelligence specialized program, and MATLAB/Labview.

(43) The input data collector 100 may comprise a microphone 112 for measuring noise of the vehicle, a vibration sensor 111 for acquiring vibration data of the vehicle, a CAN module 113, an ICP amplifier 321, an AD board 322, and a resampling unit 323 for acquiring engine speed of the vehicle.

(44) The external storage data 200 comprises a memory 201 and a cloud network 202. The memory 201 stores problematic noise regions from the noise, vibration and engine speed data.

(45) The portable device 300 includes a controller 300-1 and a problematic noise diagnosis unit 320. The controller 300-1 comprises an RPM information unit 325, a first ASCII conversion units 327, a second ASCII conversion units 328, a display unit 421, a sound reproducing unit 422 and an artificial intelligence 420.

(46) The controller 300-1 converts the data into codes and receives a diagnosis result from the artificial intelligence 420.

(47) The problematic noise diagnosis unit 320 diagnoses a problematic noise source out of the noise, vibration and engine speed based on the codes by deep learning of the artificial intelligence 420.

(48) The input data collector 100 and the problematic noise diagnosis unit 320 or the controller 300-1 or the portable device 300 can be separated from each other.

(49) An algorithm according to another exemplary embodiment of the present disclosure generally follows the following procedures.

(50) The external storage data 200 outside the vehicle is prepared.

(51) The external storage data 200 outside the vehicle is acquired and converted into a second code at the first and second ASCII conversion units 327 and 328 through resampling at the resampling unit 323.

(52) At this time, the frequency of resampling may be twice the maximum frequency of the problematic noise.

(53) In other words, as the resampling is necessary for preprocessing input data, the resampling may be performed such that the frequency thereof is twice the frequency band of the problematic noise.

(54) For example, upon diagnosing the problematic noise below 16 kHz, the frequency of resampling is set to 32 kHz.

(55) On the other hand, the external storage data 200 outside the vehicle may be transmitted to the vehicle CAN data of the CAN module 113 of the vehicle by Bluetooth.

(56) The artificial intelligence 420 diagnoses a problematic noise source out of the noise or vibration based on the second code converted by deep learning at the first and second ASCII conversion units 327 and 328.

(57) The problematic noise diagnosis unit 320 may comprise at least one of a display unit 421 or a sound generation unit 422.

(58) The artificial intelligence 420 may output the result of diagnosing the problematic noise source to the display unit 421 visually and reproduce a sound for identification in the sound generation unit 422.

(59) More specifically, the microphone 112 measures noise of the vehicle.

(60) The vibration sensor 111 acquires vibration data.

(61) The CAN module 113 acquires engine speed (RPM) of the vehicle.

(62) The controller 300-1 is connected to the microphone 112, the vibration sensor 111 and the CAN module 113 to receive respective data.

(63) The controller 300-1 converts the input data into codes and receives the diagnosis result from the artificial intelligence 420.

(64) The artificial intelligence 420 diagnoses a problematic noise source out of the noise, vibration and engine speed based on the codes by deep learning.

(65) In this case, the memory 201 may be utilized.

(66) The memory 201 of the external storage data 200 is external data outside the vehicle, which contains noise, vibration and engine speed.

(67) The memory 201 may be any one of an SD memory card, a USB memory, or any other cloud shared memory.

(68) The controller 300-1 converts the data resampled from the memory 201 into codes and receives the diagnosis result from the artificial intelligence 420.

(69) The artificial intelligence 420 can diagnose a problematic noise source out of the noise, vibration and engine speed based on the codes by deep learning.

(70) On the other hand, a bidirectional method and a gated recurrent unit (GRU) technique are applied to the deep learning of the artificial intelligence 420.

(71) The GRU refers to one of recurrent neural network (RNN) technique wherein it is possible to construct a learning model for the artificial intelligence, which follows time change, by means of the GRU.

(72) Further, an attention mechanism technique may be applied to the deep learning of the artificial intelligence 420.

(73) The attention mechanism technique can be used for improving information loss of past time data.

(74) The attention mechanism technique refers to a learning technique that imparts weights to important feature vectors in the time axis.

(75) In other words, the attention mechanism technique can establish an accurate learning model with respect to atypical noise by further imparting weights of important parts of learning data.

(76) In this case, the attention mechanism technique may be applied in combination with a bidirectional recurrent neural network (RNN) (or GRU).

(77) Further, an early stage ensemble learning technique may be applied to the deep learning of the artificial intelligence.

(78) The early stage ensemble algorithm is a weighted learning technique that emphasizes importance of data in the past time among problems of the RNN that is a superordinate concept of GRU.

(79) In other words, the early stage ensemble technique can be used for dividing weights for the initial time and weights for the final time evenly to keep information.

(80) While the present disclosure has been described with respect to the specific embodiments, it will be apparent to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the disclosure as defined in the following claims.