Research On Data Processing Technology Of Ultrasonic Pipeline Internal Inspection System

In order to prevent catastrophic failures of pipelines transporting products in petrochemical industry,nondestructive testing is very important,and the detection of these pipeline failure mainly focuses on quantifying the wall thickness thinning caused by pipeline corrosion.Ultrasonic in-pipe detection technology has a strength of being sensitive to defects,and direct and accurate detection,which has become a common technology for pipeline nondestructive testing.At present,piezoelectric ultrasonic transducer and electromagnetic ultrasonic transducer are two common methods to excite ultrasound.Among them,piezoelectric ultrasonic testing is necessary to apply coupling agent,which affects the detection efficiency.In contrast with piezoelectric,electromagnetic ultrasonic can simply produce various types of ultrasonic waves without direct contact or coupling agent,but it has the problems of low conversion efficiency and can be disrupted by noise easily.the noise reduction signal processing method is a essential part to improve the efficacy and reliability of the electromagnetic ultrasonic pipeline inner detection system.Aiming at the defect detection of electromagnetic ultrasonic pipeline inner detection system,the basic principle and method of electromagnetic ultrasonic detection were studied,and the echo signal characteristics and different defect waveform characteristics were analyzed in this paper.The criterion of algorithm processing performance was adopted,and the model of echo signal was built by gaussian echo model.For the purpose to solve the problem of poor ratio of signal to noise of electromagnetic ultrasonic system echo signal,empirical mode decomposition was selected to preprocess the echo signal.According to the feature distribution of IMF after signal decomposition,the IMF screening principle based on Euclidean distance was proposed and compared with the traditional energy way.For the purpose to find a solution to the problem that the signal still has noise after preprocessing,the preprocessing signal was denoised by singular spectrum analysis.The window size selection method based on spectrum analysis and the recombination order selection method based on hurst exponent were proposed to improve the algorithm.The significance of the proposed calculation in noise reduction and feature retention is verified by simulation.Aiming at the problem of defect feature extraction,several signal envelope methods were studied.Considering the detection accuracy and amount of calculation,Hilbert-Huang Transform in peak envelope method was selected to complete signal envelope and feature extraction.By calculating the pipe thickness and other characteristic parameters,the type and size of pipe defects are inferred and determined,and then the defect identification was accomplished.Then,the data processing system software was devised using MATLAB,and the reading and processing of data and the display of thickness value were completed through interface operation,so that the algorithm was convenient for engineering application.For the purpose of testing the processing effect of the data processing technology raised in this article some standard test blocks of large area corrosion and small area corrosion were made in the laboratory stage,the algorithmic influence on the noise reduction and feature extraction of the practical detection signal was analyzed,finally,the experimental pipeline with defects was tested.The consequence of the experiment shows that this algorithm has high detection efficiency and can distinguish defect comparatively correct.