Application Of Empirical Wavelet Transform In Extraction Of Corrosion Characteristics Of Tank Bottom

Nowadays,the consumption of oil and gas resources in the world is increasing year by year.Storage tank is one of the most effective equipment in the storage of oil and gas resources,and the common failure mode of this equipment is corrosion and crack at the bottom of the tank,the root cause is the chemical and electrochemical corrosion reaction at the bottom of the tank,and the serious consequence is oil and gas leakage.Therefore,the storage tank should be inspected regularly to ensure production safety.Acoustic emission detection method can realize online detection with high detection efficiency,and has been widely used in the detection of tank floor.In the process of acoustic emission detection,the site environment is complex,and noise will affect the accuracy and accuracy of location.It is easy to get wrong information in the process of acoustic emission signal feature extraction,so how to effectively avoid noise interference is a problem to be solved.This paper mainly studies the following aspects:(1)The principle of acoustic emission detection technology,the causes of tank corrosion and the detection method of tank floor are discussed.Wavelet threshold denoising and wavelet packet denoising are used to denoise acoustic emission signals.Through analysis,the wavelet packet decomposition denoising can get higher SNR and lower root mean square error,and the signal of wavelet packet decomposition denoising is smoother.(2)Empirical Mode Decomposition and Empirical Wavelet Transform are compared and analyzed.Empirical Wavelet Transform can effectively solve the end effect and mode mixing problems of EMD.At the same time,empirical wavelet transform can effectively divide AE signals into corresponding modal components.(3)The classification effect of support vector machine using different kernel functions is compared and analyzed,and RBF kernel function is the best.The parameter optimization method of kernel function is discussed.The calculation speed of genetic algorithm is fast,and it can carry out effective classification and recognition.(4)To the storage tank to tank bottom acoustic emission testing and magnetic flux leakage testing,acoustic emission were used to detect the data collected from the characteristic vector is calculated,using support vector machine method for storage tank bottom,the acoustic emission signal classification and recognition by 13 channels of of all kinds of corrosion and corrosion identification results based on acoustic emission signal as the main acoustic emission signal.The classification and recognition results of support vector machine were evaluated by magnetic flux leakage detection.