Research On Defect Identification Method Of Ultrasonic Nondestructive Testing Based On VMD

Ultrasonic nondestructive testing technology is widely used in the field of detection because of its advantages of fast,convenience and high efficiency of detection.In the field of ultrasonic detection,Lamb wave is often used to study the detection problems on the surface and inside of the plate.However,due to the multi-mode and dispersion effect of the Lamb wave itself,the received signal often contains multiple modes,and the amplitude of the received echo signal is easy to be missing,so it is difficult to achieve accurate location and identification of defects.Therefore,how to solve these problems and realize the identification of defect signal in plate have become the topic of this research thesis.Due to the ultrasonic Lamb wave signal has the characteristics of non-linear and non-stationary,it can not be processed by the common time-frequency domain signal method.The variational modal decomposition method is a new signal processing method which combines Wiener filtering with Hilbert transform.It is commonly used to process non-linear and non-stationary mixed signals,which can effectively separate the defect signals from the received original signals.In this paper,the defect identification method of ultrasonic nondestructive detection is studied,and the variational mode decomposition method is used to decompose the ultrasonic signal which contains defect information.The feature extraction and imaging of the processed ultrasonic signal can achieve the purpose of identifying defects in plate.The number of decomposition layers and penalty factors in the variational mode decomposition process are optimized in this paper: the number of layers in the variational mode decomposition is determined by the relationship between the K and the mean of instantaneous frequency and the size of the mutual information value.Particle swarm optimization algorithm is used to optimize the penalty factors in the variational mode decomposition process.In the process of optimization,the weight matrix of SNR,smoothness and root mean square error parameters contained in the joint characteristic parameters is obtained,so as to determine the penalty factor value when the fitness function value is maximum.The variational modal decomposition results are compared with the empirical modal decomposition results,and the variational modal decomposition method is used to divide the signal accurately.The resolution ability and the fidelity of the defect signal are better than the empirical mode decomposition method.The signal-to-noise ratio(SNR)and smoothness of the reconstructed signal are improved.The ability of the variational mode decomposition method to deal with minor defects is verified by processing the signal which contains minor defects.The results show that the variational mode decomposition method can not only separate the conventional signal with defect information,but also extract the signal with small defect.The plate is scanned along the X axis and along the Y axis to obtain defect information in the plate.The amplitude of signal defect,kurtosis and skewness are extracted from the processed signal by the variational mode decomposition method,and the plate two-dimensional imaging is carried out by full multiplication and image fusion method.The results show that compared with the original unprocessed signal,the variational mode decomposition method can improve the imaging efficiency obviously;Compared with the characteristics of kurtosis and skewness,the defect echo amplitude has better effect in imaging.The size,position and shape of the defects can be fully displayed,which can completely characterize the defect information of the inspection plate and realize the effective identification of the defects of the plate.