Study On Fault Diagnosis Method Of Acoustic Images Based On Spatial Features Of 3D Sound Field

In recent years,the development of array testing methods is maturing,acoustic imagine fault diagnosis method combining array test method and acoustic field reconstruction technology transform acoustics fault diagnosis problem into the image pattern recognition problems.It shows great potential.But the information extracted from two-dimensional source surface based on acoustic image diagnosis technology arise the characteristics of redundancy,which limit the enhance of recognition rate.In another hand,it only consider source surface acoustic image,without consider the change during the propagation of sound wave,and it also lack the information of spatial phase information.So under weak failure condition,it is difficult to recognition and diagnosis.Consider the condition of this,to improve the sound diagnostic efficiency and robustness,the grey relational clustering technology is imported and redundant information on source face sound image is optimized.Besides,a method for acoustic diagnosis of three-dimensional sound field physical space characteristics is developed,which effectively improve the sound diagnostic accuracy and stability.That further expands the engineering application of acoustic imaging technology.Array measure,acoustic imaging,near field holography,feature extraction and SVM are combined to fault diagnosis.In the one hand,the grey relational clustering technology is imported for optimization of redundant information on source face sound image.Redundant feature are reduced and diagnosis result efficiency and robustness are improved effectively by this method,and the simulation and experimental results show that this conception is correct and valid.In the other hand,a fault diagnosis method based on spatial features of 3D sound field is proposed utilizing the ideology of information mapping and fusion.First,the radiation sound field in weaker fault condition is reconstructed employing the near field acoustic holography technique,and the phase information of source image is mapped into spatial domain.So the sound pressure distribution of three-dimensional field is obtained.Then,13 sections in radiation sound field are acquired sequentially within a wavelength,and the Gabor features are extracted to build spatial feature-model of sound field for recognition and diagnosis.The simulation and experimental results show that the robustness for weaker fault states are improved effectively by the fault diagnosis method based on spatial features of 3D sound field.The engineering application of acoustic imaging technology is further expanded,and a new idea is provided for the acoustic based diagnosis technique.