Equipment Condition Monitoring Based On Spatial Distribution Characteristics Of Multiple Physical Quantities In Sound Field

There are physical quantities such as sound pressure,particle vibration velocity and vibration gradient in the sound field,and the high-order quantity of the sound field also contains rich sound field information,and the high-order characteristics of the sound field can further supplement the analysis of the sound field characteristics.Based on the new state monitoring method of sound field spatial distribution characteristics,through image processing technology,extract the texture features in the sound field spatial distribution image,and explore the spatial information hidden in the sound field distribution,but this method only uses the sound pressure information,this paper applies the high-order sound field information to the condition monitoring of the equipment through information fusion technology,and uses more comprehensive sound field information to improve the stability of the algorithm.In this paper,the principle of sound pressure decomposition is analyzed,the significance of multi-physical quantities of the sound field is obtained,and the calculation method of the vibration velocity gradient and the noise-related model of the second-order sound field are studied.By modeling the sound field of the monopole model and the composite sound source model,the near-field spatial distribution image of the sound field multiphysics is obtained,and there are obvious differences in the spatial distribution images of the sound field multiphysics of different state sound sources.Therefore,this difference can be used to identify the pattern of the sound source state,and the three image texture feature extraction methods of GLCM method,Tamura method and Gabor wavelet transform method are used to extract the features of the multi-physical spatial distribution image of the sound field,and the spatial characteristics of the space are obtained,and the two pattern recognition models of SVM and SVDD optimized based on genetic algorithm are established for the two cases of all known and unknown states of the sound source.On this basis,the information fusion technology is used to optimize the two pattern recognition models,for the SVM model,DS evidence theory is used to perform decision-level information fusion on the judgment results of each sound field quantity,and for the SVDD model,PCA principal element analysis is used to perform feature-level information fusion of multi-physical quantities.Through the information fusion of multi-physical quantities of the sound field,the erroneous judgment can be reduced,the phenomenon that the recognition accuracy of individual sound field physical quantities is too low,and the accuracy of pattern recognition is improved.Finally,the algorithm is further applied to the condition monitoring of specific structures,combined with the finite element method to model and calculate the spatial distribution of the sound field multi-physical quantities of the structure,constructs the spherical shell,simplified submersible shell and gearbox structure model and calculates the spatial distribution of the sound field multi-physical quantity,simulates the change of the structure and excitation mode of the shell and the eccentricity of the gearbox gear,and uses the spatial distribution characteristics of the sound field multi-physical quantity to achieve accurate pattern recognition of the structure in different states.This paper provides a new theoretical reference and ideas for equipment condition monitoring methods.