Study On Acoustic Field And Flaw Echo For Ultrasonic Inspection Of Shaft Parts In Coal Mine Machinery

All kinds of coal mine machinery undergo high stress for long term in moisture, dust circumstances, therefore, fatigue cracks, corrosion and other flaws will occur easily and it’s possible to result in serious accident. Bases on National Natural Science Foundation of China “Ultrasonic signal extraction and intelligent recognition for key components of the coal mining machinery”(NO:51074121), the spatial acoustic field of ultrasonic transducer, model of ultrasonic flaw echo wave and methods of flaw signal extraction are deeply studied to meet the requirements of non-disassembly inspection shaft parts in coal mine machinery.The modern methods in analytical acoustics were introduced, and the spatial impulses of circular and rectangle transducer in half-space and wedge interface were studied based on spatial impulse response. Then the transient acoustic pressure distribution in radiation space, such as acoustic field along the beam axis, acoustic field distribution projected in transversal and longitudinal directions of beam axis were simulated. To ensure the accuracy, a high sampling frequency was used to calculate the spatial pulse response which results in a large amount of data being processed, in order to improve time-consuming, a fast algorithm for impulse response calculation was proposed. Firstly, impulse response of the transducer plane were calculated with a higher sampling frequency and a smaller step size as the original data, then a lower frequency was selected to interpolation the original data, at last the critical parameters select were analyzed, this solution is an exact implementation.The process of ultrasonic wave generating by pulse, propagation in the medium and received by the transducer was analyzed. A model of the flaw echo wave was established which the received ultrasonic signal can be found by convolving of the transducer vibratory profile with the flaw pulse response. The flaw and receiving transducer were divided into small units, based on the ideas of Huygens’ principle of superposition, the received ultrasonic echo due to the superposition of the scattering echo of each flaw units. Spatial impulse response and Kirchhoff approximation were applied to establish the model of ultrasonic flaw echoes, and the vibratory profile of transducer is calculated by deconvolution the plane echo with impulse response of the plane, In order to determine the essence of the echo wave, targets of different sizes and different off-axis distance were used in simulation, the received ultrasonic echo wave is explained in terms of the plane and edge waves, however, polarity is opposite, and the amplitude of the direct wave is far greater than the edge wave.In the ultrasonic nondestructive test of shaft, the ultrasonic signal was disturbed by electronic noise, structure noise and other nonlinear and non-stationary noise. Before analyzed, the ultrasonic defect echo signal must be processed to get better results. A new noise filtering method based on empirical mode decomposition(EMD) is proposed, which the signal was reconstructed use correlation coefficient between the intrinsic mode function(IMF) components and the original signal. The instrument noise was decomposed into an independent IMF component successfully and the ultrasonic signal was reconstructed adaptivity. EMD was investigated to decompose the plan echo signal with white gaussian noise into different IMFs, however, the modal mixing was caused which makes the physical meaning of the related IMF unclear. To overcome the mode mixing drawback of the original EMD, the ensemble empirical mode decomposition(EEMD) method was applied, and the effects of critical EEMD parameter was studied, when the ratio between the noise amplitude and the standard deviation of the ultrasonic signal is 0.9, the RMSE of the reconstructed signal is the lowest.Sparse deconvolution was used to research ultrasonic signal in this thesis, to overcome the ill-posed problems, the ARMA-MM ultrasonic signal deconvolution method was proposed. ARMA model was applied to describe the transducer vibratory profile on the assumption that the ultrasonic wavelet is no-causal and non-minimum phase. Ultrasonic wavelet parameters was estimated using the Yule-Walker equations established by high-order cumulants. The minimization of cost functions defined in terms of the ?1-norm is useful for sparse deconvolution, and a computationally efficient iterative algorithm for sparse deconvolution is derived using the majorization-minimization(MM) optimization method. Experimental results showed that the position and size of the defect can be extracted from the ultrasonic time sequences.