Method of extracting, device for extracting and device for inspecting abnormal sound

Abstract

Abnormal sound desired for inspection is extracted by inputting NG signal containing the abnormal sound and OK signal not containing this abnormal sound but containing noise that is contained in the NG signal, separating the NG signal and the OK signal each into a steady component having steady property, white noise and a non-steady component having non-steady property, and extracting abnormal sound or removing noise individually from the steady component, the white noise and the non-steady component having non-steady property.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of an extraction device for abnormal sound embodying this invention.

FIG. 2 is a block diagram of the extraction filter shown in FIG. 1.

FIG. 3 is a detailed functional block diagram of the spectrum noise removing block.

FIG. 4 is a detailed functional block diagram of the noise removing block.

FIG. 5 is a detailed functional block diagram of another noise removing block.

FIG. 6 is a detailed functional block diagram of the restructuring block.

FIG. 7 is a detailed functional block diagram of another restructuring block.

FIG. 8 is a detailed functional block diagram of still another restructuring block.

FIG. 9 is a block diagram of an inspection device for abnormal sound embodying this invention.

FIG. 10 is a block diagram of the extraction filter shown in FIG. 9.

Claims

1. A method of extracting abnormal sound components from a sound, said method comprising: a first step of inputting NG signal and OK signal, said NG signal containing an abnormal sound and said OK signal not containing said abnormal sound and containing noise that is contained in said NG signal;a second step of separating said NG signal and said OK signal each into a steady component having steady property, white noise and a non-steady component having non-steady property; anda third step of extracting abnormal sound or removing noise individually from said steady component, white noise and non-steady component having non-steady property.

2. The method of claim 1 wherein said second step includes the steps of: determining order of a linear prediction model; andcalculating filter coefficients and residual signal each of said NG signal and said OK signal.

3. The method of claim 1 further comprising a fourth step of restructuring the sound after said third step.

4. The method of claim 2 further comprising a fourth step of restructuring the sound after said third step.

5. The method of claim 4 wherein said fourth step comprises the steps of: calculating filter coefficients for the sound component spectrum obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom; andcalculating an output signal from residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said filter coefficients.

6. The method of claim 3 wherein said fourth step comprises the step of converting the spectrum of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom into a time signal.

7. The method of claim 4 wherein said fourth step comprises the step of converting the spectrum of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom into a time signal.

8. The method of claim 6 wherein said fourth step further comprises the step of adding the residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said time signal.

9. The method of claim 7 wherein said fourth step further comprises the step of adding the residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said time signal.

10. A device for extracting abnormal sound, said device comprising: an input device for inputting NG signal that contains abnormal sound and OK signal that does not contain said abnormal sound but contains noise that is contained in said NG signal;an extraction filter comprising blocks; andan output device that outputs signals outputted from said extraction filter;wherein said blocks carry out a method that comprises:a first step of inputting NG signal and OK signal, said NG signal containing an abnormal sound and said OK signal not containing said abnormal sound and containing noise that is contained in said NG signal;a second step of separating said NG signal and said OK signal each into a steady component having steady property, white noise and a non-steady component having non-steady property;a third step of extracting abnormal sound or removing noise individually from said steady component, white noise and non-steady component having non-steady property; anda fourth step of restructuring the sound after said third step.

11. The device of claim 10 wherein said second step includes the steps of: determining order of a linear prediction model; andcalculating filter coefficients and residual signal each of said NG signal and said OK signal.

12. The device of claim 11 wherein said fourth step comprises the steps of: calculating filter coefficients for the sound component spectrum obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom; andcalculating an output signal from residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said filter coefficients.

13. The device of claim 10 wherein said fourth step comprises the step of converting the spectrum of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom into a time signal.

14. The device of claim 11 wherein said fourth step comprises the step of converting the spectrum of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom into a time signal.

15. The device of claim 13 wherein said fourth step further comprises the step of adding the residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said time signal.

16. The device of claim 14 wherein said fourth step further comprises the step of adding the residual signal of the sound component obtained from said NG signal by extracting the abnormal sound or removing the noise therefrom and said time signal.

17. A device for inspecting abnormal sound, said device comprising: an input device for inputting a signal of sound or vibrations obtained from a target object of inspection;an extraction filter comprising blocks;a judging device for extracting a characteristic quantity from an output signal outputted from said extraction filter and using said characteristic quantity to make a judgment regarding whether said target object is good or not good; andan output device that outputs said judgment by said judging device;wherein said input device serves to prepare said inputted signal of sound or vibrations and normal data obtained from a normal object respectively as NG signal and OK signal; andwherein said blocks carry out a method that comprises:a first step of inputting said NG signal and said OK signal to said extraction filter, said NG signal containing an abnormal sound and said OK signal not containing said abnormal sound and containing noise that is contained in said NG signal;a second step of separating said NG signal and said OK signal each into a steady component having steady property, white noise and a non-steady component having non-steady property; anda third step of extracting abnormal sound or removing noise individually from said steady component, white noise and non-steady component having non-steady property.

18. The device of claim 17 wherein said second step includes the steps of: determining order of a linear prediction model; andcalculating filter coefficients and residual signal each of said NG signal and said OK signal.