基于双谱分析的磁瓦内部缺陷音频检测方法

Acoustic Inspection of Internal Defect in Magnetic Tile Based on Bispectrum Analysis

针对目前磁瓦内部缺陷采用人工音频识别存在的问题,使用双谱分析已知内部缺陷情况的磁瓦在跌落撞击试验中产生的音频信号,其结果表明内部缺陷会影响双谱的峰值分布。根据这个规律,提出一种归一化双谱3切片分析方法用于磁瓦内部缺陷的检测。该方法首先提取归一化双谱上3个特定方向的切片;其次按照峰值分布与内部缺陷的映射关系,将每个切片划分成若干频率区域,并对各区域进行分类;然后通过对不同种类的区域最大峰值比较与计算,建立识别内部缺陷的阈值。最后,验证试验对方法的有效性进行了评估,取得了97%以上的识别正确率。试验结果表明,双谱分析在磁瓦内部缺陷音频检测中具有较好的实用性。

Internal defect detection plays a key role in quality assurance of magnetic tile. In order to solve the issue of internal defect identification of magnetic tile using human auditory,bispectrum was used to analyze the acoustic signal generated in a drop impact test of magnetic tile whose internal defect was known. The analysis indicated that internal defect can affect peak distribution of bispectrum. In accordance with this rule,a method based on three-slice analysis of normalized bispectrum was presented for internal defect detection of magnetic tile. With this method,firstly,the three slices were extracted along the specific directions of normalized bispectrum. Secondly,each slice was divided into several frequency regions and all regions were classified,according to the mapping between peak distribution and internal defect. Thirdly,by comparing and calculating the maximum peak of the dissimilar regions,the threshold value was established to distinguish internal defect. The effectiveness of this method was assessed by a verification test at last. The test result showed that the identification accuracy achieved over 97%,which demonstrated that the bispectrum analysis could be applied to the acoustic inspection for internal defect of magnetic tile.