为解决磁瓦内部缺陷较难检测的问题, 提出一种模糊聚类双谱分析方法用于其内部缺陷的无损检测。该方法以磁瓦在受到撞击时产生的声振信号作为研究对象, 利用双谱分析发现内部缺陷与双谱峰值的分布区域具有映射关系, 并且模糊聚类处理后的归一化双谱能明显地反映这一特征。根据这个规律, 通过将模糊聚类双谱的对角线切片划分为若干频段, 并计算切片指定幅值所在的频段建立内部缺陷识别规则。最后由验证试验评估该方法的可行性, 得到了92.5%以上的识别精度。试验表明：模糊聚类双谱在磁瓦内部缺陷声振检测中具有一定实用性。

In order to solve the issue of internal defect detection of magnetic tile using manual work, an analysis of fuzzy clustering bispectrum was presented for non-destructive testing of internal defect. The proposed method was based on the acoustic signal created by impacting magnetic tile. Bispectrum analysis revealed a mapping relationship between internal defect and peak distribution. Normalized bispectrum processed by a fuzzy clustering algorithm can obviously illustrate this characteristic. According to the feature, the diagonal slice of fuzzy clustering bispectrum was divided into several frequency ranges. Then identification rule was established by calculating the frequency range which contained specific amplitude of the slice. At last, the effectiveness of this method was assessed by a verification experiment and the result showed that the recognition accuracy of more than 92.5% was achieved. It was demonstrated that fuzzy clustering bispectrum can be applied to acoustic inspection for internal defect of magnetic tile.