Research On Data Processing Algorithm For Online Thickness Detection

At present,pipelines are widely used in petrochemical,industrial production,water supply and drainage industries.With the long-term use of pipelines,local corrosion,knocking and so on are very easy to occur.Once unfavorable supervision leads to leakage accidents,it will cause huge property losses.Therefore,high-precision thickness monitoring service for corroded pipelines can improve the safety of operation and reduce maintenance costs.Destructive testing and shutdown testing will cause great economic loss due to noise.Nondestructive testing is the best method.In the field of thickness measurement,Eddy current measurement and magnetic flux leakage detection are less accurate.The advantages of ultrasonic testing are obvious,mainly piezoelectric ultrasonic testing and electromagnetic ultrasonic testing.At present,piezoelectric ultrasonic thickness measurement has the highest usage rate,but it needs coupling agent,is difficult to operate at high temperature and has high cost.In this situation,uncoupled electromagnetic ultrasonic thickness measurement has attracted much attention.At present,some foreign companies have applied it to boiler detection,scale detection and thickness measurement.The method has no coupling and simple operation,but the echo conversion efficiency is low,the signal to noise ratio is very low,and the noise is difficult to extract.Accurate thickness is an important research hotspot in this paper.In order to increase the flexibility of the algorithm design,the parametric model is used to construct the ultrasonic echo model,and the relevant noise is added to it.The widely used ultrasonic model is discussed.Finally,the Gauss echo model is selected for subsequent analysis.Considering the large amount of signal data,the envelope method is proposed to reduce the computational complexity without losing the TOF signal data.Considering that the ultrasonic echo signal in this paper is a non-linear and unsteady signal,a new empirical mode decomposition and envelope extraction method based on adaptive Hilbert-Huang transform is adopted,and the noise of the most signal is reduced by combining the method of singular spectrum analysis.Aiming at the inefficiency of empirical mode decomposition and envelope extraction,the ultrasonic signal was preprocessed,including reducing random noise by repeated sampling,eliminating pulsating noise by band-pass filtering,improving wavelet threshold de-noising,greatly enhancing signal-to-noise ratio,and reducing the pressure of subsequent data processing.The method of combining MATLAB and LabVIEW is used to carry out joint simulation.The method of energy time envelope and window similarity is used to get the final thickness and achieve the expected accuracy.Finally,the front panel of the thickness measuring virtual instrument is written by LabVIEW,and the basic function of thickness measurement is realized.