针对基于振动信号故障诊断存在的接触式测量弊端及传统声学故障诊断只反映部分声学信息问题,将近场声全息技术引入故障诊断克服该缺陷.基于LabVIEW平台开发出的非接触式机械故障诊断系统,利用近场声全息重建机械声源面附近声压场,得到可视化声源的声像图；从声像图中提取反映声场空间分布的灰度共生矩阵特征,结合支持向量机模式识别实现智能故障诊断.该系统具备智能诊断测试、声源可视化、数据分析等功能,操作方便,可用性强.通过对齿轮箱故障诊断,正确率达97.3％.与传统声学诊断方法的对比,证明该系统可靠、实用.%Vibration signal based fault diagnosis has the limitation of contacting measurement, and traditional acoustic-based diagnosis (ABD) can only partly reflect acoustic information. These disadvantages can be overcome by introducing near-field acoustic holography ( NAH) to fault diagnosis. A mechanical fault diagnosis system with non-contacting measurement was developed based on Lab VIEW platform. Acoustic images were obtained by reconstructing sound pressure field near sound source surface with NAH technology. Gray level co-occurrence matrix features reflecting spatial distribution of sound field were extracted from acoustic images. Combining with support vector machines, the intelligent fault diagnosis was implemented. The system has some key functions, such as fault diagnosis, sound source visualization, and data analysis. The gearbox faults were diagnosed using the proposed system with diagnostic accuracy of 97. 3% , which is better than the result by traditional ABD. The reliability and practicality was demonstrated by a practical application.
展开▼
