Research On Structure Damage Identification System Of Steel Pipeline Based On Piezoelectric Guided Wave

With the development of urbanization and the establishment of green and low-carbon circular development economic system,as an important part of China's economic construction,the scale of pipe network system is expanding,as well,which has brought great impact on the structural health monitoring of pipeline system.As is known to all that the pipeline failure events is occurrent frequently in recent years,more and more people has realized the important of research on pipeline structure health monitoring.To develop an efficient,accurate and widely used technology has become the common goal of pipeline structure health monitoring.There are many damage forms of pipeline structure,including cracks,scratches,holes,corrosion and so on.Different forms of damage require different maintenance strategies.At present,although a lot of theoretical and experimental researches have been carried out on the ultrasonic guided wave defects of pipeline structure defects at home and abroad,most research are mainly focused on detecting and locating for defects,while the research on the identification of different types of defects is relatively less.To solve this problem,this paper combines the pattern recognition technology with the ultrasonic guided wave detection experimental system to study the identification and location of different kinds of pipeline damage.This paper is aiming at the identification and location of the structural damage of the process pipeline,the 304 stainless steel pipelines with an outer diameter of 76 mm and a wall thickness of 3mm is taken as the research object,and a damage detection experimental platform based on piezoelectric ultrasonic guided wave is built.Firstly,a set of experimental schemes is established by fully studying the experimental phenomena.Then,the defect feature extraction and intelligent damage identification are realized by combining the noise reduction algorithm of the variational mode decomposition and the depth neural network model based on the multi-layer perceptron.Finally,the method of pulse echo is used to locate the defect.The specific work and main conclusions of this paper include:(1)An experimental platform of damage detection based on piezoelectric ultrasonic guided wave is built.Combined with the guided wave theory,the experimental phenomena are fully studied,and the number of cycles,frequency of the excitation signal and the arrangement of the driver are determined in the experiment.The experimental results show that there are some differences among the three testing results of flawless tube,20 mm cracked tube and 3 mm through hole tube,and the echo signal of the defect can be observed and the defect location can be realized in the detection signal of 20 mm crack defect tube.(2)The noise reduction method and feature extraction method of guided wave signal are studied.The feature extraction method of signal in time and frequency domain based on the variational mode decomposition algorithm is proposed.It is found that compared with wavelet decomposition and empirical mode decomposition,VMD can better retain the original excitation signal and enlarge the time-domain and frequency-domain features of defects.(3)Based on the difference of time-frequency characteristics of different kinds of damaged pipeline signals,a deep neural network structure damage identification method based on multilayer perceptron is proposed.Through the model comparison experiment,it is found that the depth neural network structure damage identification method based on multi-layer perceptron is superior to the traditional BP neural network model in the accuracy rate,model training efficiency and model stability.A damage location method based on pulse echo theory is proposed.The result shows that the location result of this method can be accurate to within 0.1 meters.