Research On Sparse Sampling Method Of Ultrasonic Signal Based On Exponential-Reproducing-Kernel

Ultrasonic testing is one of the important detection methods in the field of nondestructive testing,and it has been widely applied in the industrial and medical fields.However,because of the high frequency of the ultrasonic signal,if the traditional sampling method is adopted,the sampling frequency will be above tens of MHz.In addition,higher frequency ultrasound is often required in order to improve the quantization precision of the detection and the ultrasonic detection is gradually developed into the direction of multi-sensor like array,result from which,higher requirements for signal acquisition,transmission and storage are put forward for the mass data.In order to solve this problem,some scholars have proposed the FRI(Finite Rate of Innovation)sampling method and applied it to the sparse sampling of ultrasonic signals,which make an effective reduction in the amount of data collection.However,the method is still in the stage of theoretical research and simulation,and there are few related hardware implementations which still have some limitations in practicality and generality.Therefore,based on the FRI sampling theory,the theory and hardware implementation method of sparse sampling for ultrasonic signal are studied in this thesisIn this thesis,the existing FRI sparse sampling theory and the reconstruction framework are compared and analyzed,and Exponential-Reproducing-Kernel is chosen as the sampling kernel based on which to carry out the research on sparse sampling theory and hardware implementation method of ultrasonic signals.On the theory,a framework for sparse sampling and reconstruction of ultrasonic signals based on ERK(Exponential-Reproducing-Kernel)is established to avoid the hardware realization complexity of signal periodic extension,and a new method of ERK sparse sampling with arbitrary pulse position is proposed,which makes up for the existing ERK sparse sampling.On the hardware,a hardware implementation framework of ERK is set up,by which,the ERK-based sparse sampling hardware system for the ultrasonic signal is designed,which physical realize the sparse sampling of the ultrasonic signal with pulse at any position.Then the actual test is carried out on the pipeline defect detection platform,and the sparse sampling data is obtained,by which,the amount of data collection is greatly reduced and the parameters of the ultrasonic echo signal are accurately obtained by estimating the parameters from the measured data.The validity of the sparse sampling method and its hardware system is verified.The main research work and conclusions are as follows:1)On the basis of detailed description and analysis of the FRI sparse sampling theory,the characteristics of different sampling frames and sampling kernels are compared and analyzed.Basing on the existing sparse sampling framework,an improved ERK sparse sampling theory framework for ultrasonic signals is established,which provides theoretical support for hardware implementation of the sparse sampling method.2)Point out the problem that the position of pulse echo is limited in time domain in the existing ERK sampling method,and the reason for this problem is analyzed.On this basis,a new ERK sparse sampling method is proposed.This method can realize the sampling and reconstruction of the signal with arbitrary pulse echo position without increasing the number of sampling points.The validity of the sampling method is verified by the simulation test,and the adaptation range of the sparse sampling method is further expanded.3)Through the analysis of the performance structure of ERK,an ERK hardware implementation framework based on the combination of Analog-Rational-Kernel and Digital-Rational-Kernel is proposed,and both the parameter constraint conditions and parameter selection methods under this implementation framework are analyzed.Hardware circuit of the Analog-Rational-Kernel is designed and produced,and the ERK-based physical sparse sampling of the ultrasonic signal is realized with the corresponding software and hardware design,obtaining the sparse sampling data of the ultrasonic signal.4)A test platform of pipeline defect detection is set up,in the test,the sampling rate of 160 KHz is used to sample the ultrasonic signal of center frequency 5MHz.The sampling rate is only 1.6% of the conventional Nyquist sampling(the corresponding Nyquist sampling rate is 10MHz),and the characteristic parameters of the ultrasonic echo signal are reconstructed accurately,by which the sampling rate and the data amount has been effectively reduced.