作者：赵越, 殷鸣, 黄沁元, 殷国富, 谢罗峰

作者单位：四川大学制造科学与工程学院, 四川 成都 610065

关键词：音频检测;小波包;BP神经网络;磁瓦;内部缺陷

摘要：

针对工业生产中提高磁瓦内部缺陷检测效率、降低误检率和漏检率的实际需求, 提出一种小波包变换(wavelet packet transform, WPT)结合人工神经网络(artificial neural network, ANN)的磁瓦内部缺陷检测方法。通过采集合格和缺陷磁瓦撞击金属块产生的声音信号, 用小波包分解与重构, 筛选并提取特定频段信号的能量作为特征信息, 输入BP神经网络并训练, 使BP网络具有磁瓦内部缺陷检测的功能, 试验证明该方法准确率达到98%以上。结果表明:小波包-神经网络方法(WPT-ANN)检测速度快、可靠性高、适应性强, 为高效、准确地进行磁瓦内部缺陷检测提供有效的技术手段。

Acoustic impact testing of magnetic tile internal defects based on wavelet packet transform and artificial neural network

ZHAO Yue, YIN Ming, HUANG Qinyuan, YIN Guofu, XIE Luofeng

School of Manufacturing Science and Engineering, Sichuan University, Chengdu 610065, China

Abstract: The paper has proposed a novel approach to examine the internal cracks of magnetic tiles based on wavelet packet transform (WPT) and artificial neural network (ANN), with the purpose of improving the detection efficiency and reducing the false detection ratio and the missing detection ratio in industrial production. The acoustical signals, separately generated from the collision of a qualified and a flawed magnetic tile with a metal block, were collected and then decomposed and reconstructed by WPT. The power of the signals at specific frequency bands was filtered and extracted as the feature vector of the BP neutral network to detect the quality of the magnetic tiles. The experiment proves that the identification accuracy of this method has been up to above 98%. The experimental results demonstrate that the proposed method is fast, reliable, adaptable, accurate and efficient.

Keywords: acoustic impact testing;wavelet packet transform;BP neural network;magnetic tile;internal defect

2015, 41(6): 81-85  收稿日期: 2015-1-13;收到修改稿日期: 2015-2-25

基金项目: 四川省科技支撑计划项目(2014GZX0001, 2014Z0119)

作者简介: 赵越(1991-),男,内蒙古赤峰市人,硕士研究生,专业方向为光机电一体化技术及设备。

参考文献

[1] Li X, Jiang H, Yin G. Detection of surface crack defects on ferrite magnetic tile[J]. NDT & E International, 2014(62):6-13.
[2] Wu J, Liu C. An expert system for fault diagnosis in internal combustion engines using wavelet packet transform and neural network[J]. Expert Systems with Applications, 2009, 36(3):4278-4286.
[3] 司景萍, 郭丽娜, 牛家骅. 基于小波包神经网络的发动机振动信号分析[J]. 振动、测试与诊断, 2013, 33(5):763-769.
[4] Wang G, Wang Z, Qin X, et al. Accurate diagnosis of rolling bearing based on wavelet packet and RBF neural networks[J]. Beijing Keji Daxue Xuebao/Journal of University of Science and Technology Beijing, 2004, 26(2):184-187.
[5] Gokmen G. The defect detection in glass materials by using discrete wavelet packet transform and artificial neural network[J]. Journal of Vibroengineering, 2014, 16(3):1434-1443.
[6] Avci E, Turkoglu I, Poyraz M. Intelligent target recognition based on wavelet packet neural network[J]. Expert Systems with Applications, 2005, 29(1):175-182.
[7] Bahoura M, Simard Y. Blue whale calls classification using short-time Fourier and wavelet packet transforms and artificial neural network[J]. Digital Signal Processing: A Review Journal, 2010, 20(4):1256-1263.
[8] Sengur A, Turkoglu I, Ince M C. Wavelet packet neural networks for texture classification[J]. Expert Systems with Applications, 2007, 32(2):527-533.
[9] 宋知用. Matlab在语音信号分析与合成中的应用[M]. 北京:北京航空航天大学出版社, 2013:44-45.
[10] Zamani Hosseinabadi H, Amirfattahi R, Nazari B, et al. GUW-based structural damage detection using WPT statistical features and multiclass SVM[J]. Applied Acoustics, 2014(86):59-70.
[11] 陈恬, 孙健国. 粗糙集与神经网络在航空发动机气路故障诊断中的应用[J]. 航空动力学报, 2006, 21(1):207-212.
[12] Bin G F, Gao J J, Li X J, et al. Early fault diagnosis of rotating machinery based on wavelet packets-Empirical mode decomposition feature extraction and neural network[J]. Mechanical Systems and Signal Processing, 2012, 27(1):696-711.
[13] 顾文龙, 胡业林, 郑晓亮. 基于小波包神经网络的电机故障诊断分析与研究[J]. 煤矿机械, 2011, 32(9):263-265.