基于光纤光栅和支持向量机的声发射定位系统
Acoustic Emission Localization System Based on FBG Sensors and Support Vector Machine

利用光纤光栅传感器和边缘滤波原理构建传感系统，结合小波分解与重构和支持向量机算法，对铝合金板声发射定位进行了研究。根据划分区域进行声发射实验，探索声发射源所在区域与信号特征之间的关系。在对声发射信号进行小波分解的基础上，使用近似系数和细节系数进行[JP2]重构，并对重构后的各信号计算其振荡能量作为信号特征，进行声发射区域识别。以重构信号的振荡能量作为输入、声发射区域位置类别作为输出构建支持向量机多分类模型，实现了声发射区域定位识别。实验结果表明，在400 mm×400 mm×2 mm的铝合金板上对36个测试样本进行了多次声发射区域定位识别，在180次模拟实验中实现了176次声发射区域准确定位，正确率达到97.78%，声发射区域识别精度为30 mm×30 mm。该研究结果为机械结构的声发射区域定位检测提供了有效方法。
We constructed an acoustic emission (AE) localization system for aluminum alloy plate using fiber Bragg grating (FBG) sensor network and edge filtering, then used the wavelet transforms and support vector classifiers algorithm to identify acoustic emission localization. We explored the relationship between the AE region and signal characteristic according to the acoustic emission test on the divided area. On the basis of signal wavelet decomposition, we proposed the AE region identification method, which is based on the vibration energy calculated by a reconstructed signal using approximation coefficients and detail coefficients. Then, we built a multi-classification support vector machine whose input was vibration energy of the reconstructed signal and output was the AE region. The experimental results showed that for a 180 AE area identification test on a 400 mm×400mm×2mm aluminum alloy plate, a 176 AE region identification test was entirely accurate, which indicated accuracy at 97.78% and on a 30mm×30mm regional location.