Fault Feature Extraction Technique Based On Patch Near-Field Acoustical Holography

As the result of vibration emission in air, machine vibration and sound signals carry affluent information about the working condition of machine, and it can be used to make mechanical fault diagnosis. However, the fault diagnosis technique based on vibration signals has its restrictions. On the other hand, machine sound signal also carries some useful information about the working condition of machine, and can partly take place of vibration signal and to be used for mechanical fault diagnosis. Furthermore, machine sound signal has the advantage of non-contacting measurement. To implement the fault diagnosis technique based on sound signal, it should take into account of the characters of mechanical noise and make deep research into the technique of acoustical feature extraction. So, the extracted feature can describe mechanical equipment'condition well. Traditional acoustical feature extraction method can only describe the change law of machine fault feature on time, frequency, but it can not reveal the changing information of fault features according to the locations of sound sources. For more efficiently perform mechanical fault diagnosis using sound signal, a fault feature extraction based on patch near-field acoustical holography (PNAH) is presented in this work. This technique use an array consisted of a small number of microphones to acquire partial acoustic quantities, and data extrapolation (or data interpolation) technique based on wave superposition method (WSM) is adopted to reconstruct the acoustic parameters including sound pressure, velocity and sound intensity at every point in 3D sound field. Furthermore, the hologram including the number of sources, the location of sources and the intensity of sources is obtained, so sound source identification and source field prediction can be realized. Meantime, comparing the hologram under the condition of normal and abnormal status, the change of sound source feature can be identified in the specific position, and the specific fault can be found when considering the characteristic frequency and the feature parameter of the equipment. Near-field acoustical holography demands that one side of holographic surface must be free. In order to overcome this limitation, a sound field separation technique based on WSM is proposed when the sound pressures are obtained using single holographic surface. When there are sound sources at both side of holographic surface, this technique can separated each acoustic quantity from the mixed signal of the holographic surface. Then, the hologram of the target sound source can be reconstructed from the separated acoustic quantity, and the fault feature can be extracted. The main contents of this dissertation can be summarized as follows. Firstly, the background of fault diagnosis will be discussed. The research history of machinery fault diagnosis technique and acoustical fault diagnosis technique will be summarized. The advantage and disadvantage of every sound source identification methods are analyzed and compared detaily. With the aim of fault feature extraction using acoustical feature methods, questions to be solved by this work is proposed, and a foundation of this work is constructed.After that, the basic theory and reason of sound radiation from vibrating structure are introduced. And the mechanical sound radiation problems are described from the viewpoint of mathematics. And the near-field acoustical holography (NAH) algorithm is deduced and discretized. Also, some filters are discussed in wave number domain. Through numerical simulation, the results show that NAH can accurately identify sound sources in some extent. But it also shows that it causes warp-round errors and windowing effect in the calculations.The theory of WSM is investigated. The correlative theories and implement ways of WSM are discussed. The wave superposition formulation is deduced with principle of wave superposition. And the non-uniqueness of the solution is discussed. Since the sound field reconstruction belongs to the ill-problem and its solution is not stable, so the ill-posed question and the regularization method are discussed when discretizing the wave superposition integral formula. Through numerical simulation, the following factors that influence the reconstructed accuracy are chosen to be discussed: holographic surface, reconstruction surface, equivalent source surface, and measurement errors, etc. And some guidelines will be found in order to improve the reconstructed accuracy. The traditional NAH demands that one side of holographic surface must be free, and this seriously restricts the application range of acoustical holography technique in the actual engineering. In order to overcome this limitation, a sound field separation technique based on single holographic surface and WSM is proposed when there are sound sources at both side of the holographic surface. Through numerical simulation and experiment, the feasibility and accuracy of this technique are validated. Based on the sound field separation technique, the sound field produced by sound sources at both side of the holographic surface can be studied, and this have promoting effect for the popularization and application of acoustical holography.Conventional NAH is based on discrete spatial Fourier transform. To avoid truncation errors caused by the finite measurement aperture, the measurement must be implemented over a large area. However, it is impractical for large structure when the reconstruction of sound field is required only on a partial structure surface. In order to resolve the problem, a data extrapolation technique based on WSM is proposed. The theory and implement process about this technique are analyzed, including data extrapolation and sound field reconstruction. Since WSM has a precondition that it should be told where to collocate the equivalent sources, the location of sound sources is obtained according to beamforming method and a serial of equivalent sources are collocated around these sound sources. Through numerical simulation, the validity and accuracy of this technique are proved; also the results demonstrate the superiority of this technique in comparison to traditional NAH. Meantime, the frequency range of NAH based on spatial Fourier transform is limited by the special dimension of holographic surface, and this is low frequency threshold and high frequency threshold. As for large scale objects at high frequency, it will need a large number of microphones and the testing cost will increase. So a data interpolation method based on WSM is presented. Under the premise of up to same spatial resolution, this method can decrease the number of measurement points and save the workload of measurement. At last, the accuracy and feasibility of this method are validated through numerical simulation, and the results show that this method has great practical value.After that, experiments are done to evaluate feasibility and accuracy of the techniques presented by this work: data extrapolation technique, data interpolation technique, and the acoustical fault feature extraction technique base on it. That will make a basis for its application in the industrial field. The present hardware of the vibration laboratory is introduced. A set of microphone array and data acquisitions system are designed and implemented. The principle of the experiment is described. On basis of that, experiments are performed in a semi-anechoic chamber. The sound source models are made up of sound boxes and motor. Sound source identification and fault feature extraction are performed on this sound source model. After acquisition of the sound pressure data, the experimental results are analyzed. The efficiency and accuracy of the proposed techniques are proved by experimental results.Conclusions are given at last. The innovations are summarized. At the same time, some advices are given for the future research for the fault feature extraction based on acoustical holography. Also, remarks are given on some problems.