融合声场法向变化的双面声像故障诊断方法研究

Acoustic fault diagnosis technology based on dual audio-visual images

声像诊断技术把声诊断问题转化为了图像识别问题,改善了传统的基于单点测试的声诊断鲁棒性,但由于单个声像忽略了三维声场的空间变化信息,在弱故障工况下诊断困难。针对上述问题,基于声场三维特性和信息融合思想,提出了一种基于双面声像模式识别的故障诊断方法。首先基于近场声全息技术构建融合源面声像、全息面声像和两者差值声像的双面声像模型,然后提取Gabor小波纹理特征,并基于随机森林特征选择算法进行特征降维,构建有效声场特征模型进行状态诊断识别。仿真和实验结果表明,基于双面声像模式识别的故障诊断技术是有效可行的,能有效改善弱故障工况的诊断鲁棒性,进一步拓展和改善了基于阵列测量的声像诊断技术。

The near field acoustic holography(NAH)-based fault diagnosis method transforms acoustic-based diagnosis(ABD)problem into image identification.It improves the robustness of traditional ABD technique using single-channel test analysis.But due to neglecting sound spatial change information,the acoustic diagnostic technique is difficult to be implemented in weak conditions.Aiming at the problems above,an acoustic fault diagnosis technique based on dual audio-visual images is proposed,in which the physical properties of sound field and the information fusion are adopted.The dual audio-visual model including source image,holographic sound image and their difference image is constructed.Then Gabor wavelet texture features are extracted from the dual audio-visual model and the random forest feature selection algorithm is implemented to construct an effective sound field feature model for recognition and diagnosis.The simulation and experimental results show that the robustness for the weak fault states are improved effectively by the new fault diagnosis method based on dual audio-visual images.The engineering application of acoustic imaging technology based on array measurement is further expanded,and a new idea is provided for ABD technique.