Sub-Nyquist Sampling And Reconstruction Of Pipeline Defect Ultrasonic Testing Signal

Pipeline transportation is a common method of oil and gas transportation with the advantages of low cost, large transport capacity, safety and reliability. But with the development and popularization of the pipeline, regular test and maintenance of pipeline becomes more and more important. Pipeline defect ultrasonic internal testing is an important nondestructive testing method for flaw in service inspection. However, large amount of sampling data in pipeline ultrasonic testing leads to storage and online processing problems. This thesis focuses on research sub-Nyquist sampling methods and signal reconstruction algorithms to reduce sampling data and reconstruct signal from the sparse sampling data.Based on analyzing the traditional Nyquist sampling theory, both the periodically non-uniform sampling theory and wavelet sampling theory which breakthrough Nyquist sampling frequency limitation have been researched. The effects of reducing acquisition data through these two sub-Nyquist sampling methods are analyzed by sampling and reconstructing ultrasonic signals on simulation ultrasonic signal and real testing signal. Signal relative reconstruction error and peak arrival time error are introduced for performance evaluation. The experimental results show that wavelet sampling method is better than the periodic non-uniform sampling method for ultrasonic signal sparse sampling.Although the periodic non-uniform sampling and wavelet sampling methods can reducing the sampling data, but the compression ratios of both methods are not high. To further improve the compression ratio in sampling process, analog-to-information convertor(AIC) theory based sampling theory has been applied and studied. The performance of this sampling method is verified by pipeline experiments; and the results indicate that the compression ratio is improved than above two methods.According to the modulated wideband converter(MWC) sampling theory and ultrasonic echo signal characteristics, a modulated single-band multi-channel converter(MSMC) compressive sampling framework is proposed, which is applicable to ultrasonic echo signal sampling. The signal reconstruction performances at different compression ratios and noise level have been studied. For the purpose of reducing the reconstruction noise, wavelet denoising method is introduced. Experimental results show that MSMC framework can obtain higher compression ratio and better reconstructed performance compared with AIC framework.