1. Non contact acquisition of sonic emissions of bearings Assoc. Prof.: Kiril M. Alexiev, Petia D. Koprinkova-Hristova; Dr.: Vladislav V. Ivanov, Volodymyr V. Kudriashov, Iurii D. Chyrka AComIn Technology Transfer Workshop on Advanced Techniques in NonDestructive Testing Sozopol, June 18-19, 2015

2. Outline The Acoustic Camera overview and resolution enhancement Experiment description Comparison of spectra estimates from single microphone and from the focalized Acoustic Camera Obtained spectra estimates for different bearings 2

3. Acoustic Camera Applications: Acoustic Imaging and Signal Analysis Pictures from WWW 3

4. The Acoustic Camera 4 Software list: a) Acoustic Test Consultant - Type 7761; b) Beamforming - Type 8608; c) FFT Analysis - Type 7770; d) Time Data Recorder - Type 7708. Manufacturer: Brüel & Kjær (Sound and Vibration Measurement A/S) PULSE LabShop Customized Solution Version 17.1.2 Array Acoustics Post-processing ver.: 17.1.2.308 Frequency range: from 10Hz to 20kHz Wavelength range: from 34.3m to 17.15mm

5. Criteria: Half power (-3dB) 55 Trg. 2Trg. 1 Resolution Enhancement

6. 6 B&K “Array Acoustics Post-processing”  1 responseCapon mod.  2 responses dB Center frequency f c =10 kHz. Frequency bandwidth  f ≈ 2.3 kHz. f c /  f ≈23% Range ~0.8 m. Spacing between centers of the speakers ~0.1 m BeforeAfter (Obtained result) Trg. 1Trg. 2Trg.

7. 77 Beamforming based-on modified Capon algorithm dB Center frequency f c =5 kHz. Frequency bandwidth  f =0.5 kHz. f c /  f =10% Range ~0.75 m; Spacing ~0.14 m “Delay and Sum” BeamformingCapon mod. BeforeAfter (Obtained result) Trg. 1Trg. 2Trg.

8. Photo 8

9. Spectra Estimates 9 Parameters for SKF bearing SKF ”type” 6205. Number of rolling elements N9 Rolling element diameter B, mm7.938 Pitch diameter P, mm39.04 Contact angle , deg. 0 Rotational speed F, rpmUnknown. Slightly variable. SKF old Opened

10. 10 Verified with SKF “Bearing Frequencies Calculator”. Bearing SKF ”type” 6205.  Parameters on  Rotational speed F, rpm & Shaft speed frequency, Hz Min: 1820 & 30.3(3)Max: 1830 & 30.5 Fundamental train frequency FTF, Hz 12.082812.1492 Ball pass frequency BSF, Hz 71.507671.9005 Ball pass frequency outer race BPFO, Hz 108.7455109.343 Ball pass frequency inner race BPFI, Hz 164.2546165.157 Rolling element defect frequency, Hz 143.015143.801 Bearing Fundamental Frequencies SKF old Opened

11. The first bearing band 11 SKF old Opened Acquisition time, T = 0.25, [s] Frequency resolution,  f = 4, [Hz] Quantity of autospectra averages = 100 Good sample emission is more powerful Bad sample emission is more powerful

12. Spectra estimates 12 Acquisition time, T = 0.25, [s] Frequency resolution 1,  f 1 = 4, [Hz] No averaging Acquisition time is T Frequency resolution 2,  f 2 = 256, [Hz] Quantity of autospectra averages = 64

13. Non-stationary signals 13 The considered spectra difference is at higher frequency range. Thus, such peaks were not removed, yet.

14. Spatial Filtering 14

15. Enhancement of Spectra Difference 15 The difference is not getting worse due to application of the microphone array Beamwidth ~60 o Beamwidth >~120 o

16. The Generated Acoustic Map 1 16 Lower Signal-to-noise ratio - 55 dB Center frequency, f С =10, [kHz] Bandwidth  f = 1, [kHz]

17. The Generated Acoustic Map 2 17 Higher Signal-to-noise ratio - 55 dB up grow, PSF Center frequency, f С =10, [kHz] Bandwidth  f = 1, [kHz]

18. Spectra Estimates 18 SKF old Opened Hereinafter: Acquisition time, T = 0.25, [s] Frequency resolution,  f = 4, [Hz] Quantity of autospectra averages = 100

19. Spectra Estimates 19 SKF old Opened The difference is much lower than in previous Measurements Session due to greasing

20. Spectra Estimates 20 SKF old Opened At the room, the background varies at low freq.

21. Spectra Estimates 21 2. WBF Good Two “same” spectra About 20 dB difference

22. Spectra Estimates 22 3. NSK

23. Spectra Estimates 23 4. KBS

24. Spectra Estimates 24 5. HF

25. Spectra Estimates 25 6. CNR

26. Spectra Estimates 26 7. VMF

27. Spectra Estimates 27 8. SKF

28. Spectra Estimates 28 8. SKF

29. 29 Main goal is to find collaboration with industry/business. Project AComIn "Advanced Computing for Innovation“ FP7 Capacity Programme. Host – Institute of Information and Communication Technologies at the Bulgarian Academy of Sciences.

30. Acoustic Camera Applications 1. Noise pollution Airport noise Urban/Street noise Instrument noise 2. Noise identification (Find the source of specific noise) Sound quality analysis Data/music recording Multimedia product analysis 3. Diagnostic approaches Spectral analysis Time-frequency analysis Trend detection in sound intensity signals Noise intensity analysis Shock response analysis etc, including post-processing in third-party software 4. Occupational health Noise exposure Hearing protection Human vibration Noise reduction Factory hall acoustic 5. Military/security applications Noise barrier detectors Noise localization Noise recognition 6. Scientific tool for research in Beamforming Random antenna array development Acoustic signal analysis Acoustic holography Signal processing/filtering etc. 7. Educational Useful for experiments/demonstrations 30

31. Conclusions Frequency resolution of the Acoustic Camera is modified from 1/3 octave to 4 Hz. Comparison of the spectra shows opportunity to detect difference to good bearings. Non-contact diagnosis is implemented. Future Plan: Automatic detection of defects (using the spectra) will be implemented for priori unknown background noise. 31