Feature extraction of wood-hole defects using wavelet-based ultrasonic testing

Abstract

The primary bottleneck to extracting wood defects during ultrasonic testing is the accuracy of identifying the wood defects. The wavelet energy moment was used to extract defect features of artificial wood holes drilled into 120 elm samples that differed in the number of holes to verify the validity of the method. Wavelet energy moment can reflect the distribution of energy along the time axis and the amount of energy in each frequency band, which can effectively extract the energy distribution characteristics of signals in each frequency band; therefore, wavelet energy moment can replace the wavelet frequency band energy and constitute wood defect feature vectors. A principal component analysis was used to normalize and reduce the dimension of the feature vectors. A total of 16 principal component features were then obtained, which can effectively extract the defect features of the different number of holes in the elm samples.

Keywords

Wood / Performance extraction / Wavelet energy-moment / Principal component analysis

References

[1] Bamford M, Batsale JC, Fudym O. Nodal and modal strategies for longitudinal thermal diffusivity profile estimation: application to the non destructive evaluation of SiC/SiC composites under uniaxial tensile tests. Infrared Phys Technol, 2009, 52: 1-13.

[2] Oh JK, Kim CK, Hong JP, Lee JJ. Improvement of robustness in ultrasonic attenuation spectroscopy for detecting internal insect damage in wood member of cultural heritage. J Wood Sci, 2015, 61(2): 136-142.

[3] Qi W, Wang LH. Identifying the patterns of defects in timber using ultrasonic test based on wavelet neural networks. Sci Silvae Sin, 2006, 42(8): 63-68.

[4] Sfarra S, Theodorakeas P, Avdelidis NP, Koui M. Thermographic, ultrasonic and optical methods: a new dimension in veneered wood diagnostics. Russ J Nondestr Test, 2013, 49(4): 234-250.

[5] Sun JP, Wang FH, Zhu XD. Application of wavelet-neural network in defect location non-destructive testing of MDF. Chin J Sci Instrum, 2008, 29(05): 955-958.

[6] Sun JP, Hu YC, Wang FH. Study on quantitative nondestructive test of wood defects based on intelligent technology. Chin J Sci Instrum, 2013, 34(09): 1955-1960.

[7] Wang EC, Qiu XC. New method of lumber recognition using improved C-V model and wavelet transform. Comput Eng Appl, 2011, 47(8): 211-214.

[8] Wang LH, Yang HM. Application of wavelet packet analysis and BP ANN in diagnosing the hole defects in Acer mono wood using ultrasonic quantitative testing. J Beijing For Univ, 2007, 29(02): 128-132.

[9] Wang FH, Zhu X, Sun JP. Applications of wavelet analysis in the nondestructive test of medium density fiberboard. Sci Silvae Sin, 2004, 42(10): 91-94.

[10] Wang BX, Liu WF, Liu JN, Cui YY, Luo XZ. Feature based method for classifying and detecting ultrasonic signals. J Tsinghua Univ, 2012, 52(7): 941-945.

[11] Yang HM, Yu L, Wang LH. Effect of moisture content on the ultrasonic acoustic properties of wood. J For Res, 2015, 26(3): 753-757.

[12] Yu GX, Zhang AZ, Shi BZ. Detection of timber decay by stress wave frequency spectrum. J Northeast For Univ, 2007, 35(10): 22-25.

[13] Zhang YR, Fu L, Zhou XQ. Near infrared detection of wheat water content based on bp neural network. J Henan Univ Technol, 2013, 34(1): 17-20.

[14] Zhao D, Zhu HJ. Study on acoustic emission of wood defects based on wavelet packet analysis. Comput Eng Appl, 2010, 46(11): 220-222.

[15] Zhou JM, Sun K, Li B, Li P, Xu QY. Simulation and Localization of Ultrasonic Defect Detection Based on Array Probe. J East China Jiaotong Univ, 2015, 32(5): 105-109.

[16] Zhu HJ (2007) Acoustic emission testing for wood materials based on wavelet neural network. In: Dissertation of Beijing Forestry University, Beijing: Beijing Forestry University