Research On Ultrasonic Phased Array Data Compression Technology And Defect Detection Technology

Ultrasonic phased array detection has been widely used in the field of nondestructive testing because of its advantages of safe use and flexible beam.At present,the complexity and volume of workpiece detection has increased in ultrasonic phased array detection,and all the detection data of workpiece need to be stored from the beginning to the end of its life cycle detection.As a result,the problem of large amount of detection data is generated and the difficulty of accurate extraction of defect features is increased.According to the above requirements,ultrasonic phased array signal feature extraction and signal data compression were studied.In the aspect of ultrasonic phased array feature extraction,sparse feature extraction method was used to process ultrasonic phased array defect signal in this article.Firstly,according to the characteristics of ultrasonic echo,Gabor,predefined and K-SVD dictionaries were constructed to represent ultrasonic echo,and reconstruction errors of three greedy algorithms were compared.Secondly,different sparse features were extracted from different dictionaries,and BP neural network was used as a classifier to identify defects.In the aspect of ultrasonic phased array data compression,optimal measurement matrix compressive sensing method was used to compress ultrasonic phased array signal in this article.The optimal measurement matrix was constructed by optimizing correlation and column independence.The compressed A-Scan signal was obtained by beamforming single array element ultrasonic signal compression sensing,which improves the traditional A-Scan signal compression sensing method.Carbon fiber reinforced plastic was used as specimen to study the method of this article.In the aspect of ultrasonic phased array feature extraction,experimental results show that different types of sparse features could accurately obtain information such as defect location and type.Under BP neural network,the sparse features have satisfactory recognition and anti-noise effect compared with the traditional time-domain features and wavelet features.In the aspect of ultrasonic phased array signal compression,experimental results show that the optimal measurement matrix has better reconstruction effect compared with traditional matrices.Under 60% compression rate,the reconstruction error of single array element ultrasonic signal compression sensing beamforming A-Scan is only 2.978%,and the peak signal-to-noise ratio of B-Scan image is 43.1093.Under the same compression rate,the improved method has better reconstruction effect.