| In engineering,steel structures are widely used in bridges,buildings etc.Welding is the most common way of connecting steel structures.During the welding process,residual stresses are generated as a result of the rapid expansion and contraction of the steel due to the rapid change in temperature.The residual stresses formed after welding will not only reduce the stiffness and stability of the structure,but also have a greater impact on the structural strength and fatigue strength of the steel members.Therefore,it is important that the residual stresses in welding are detected with high accuracy and efficiency.Magnetic memory technology as a new non-destructive testing means,because of its superiority in equipment and other aspects and attention.In this paper,the relationship between welding residual stress and leakage signal strength is explored by means of finite element simulation,experiment and data processing based on the principle of magnetic memory technology detection and force-magnetic coupling,the main contents of this thesis are as follows:Firstly,the numerical simulation of ordinary welded steel members using ANSYS software,the temperature and stress fields were analysed separately,and the distribution pattern and size of residual stresses in steel butt welds were derived using the thermalforce coupling in finite elements.Then COMSOL finite element simulation software as the basis for the construction of a simulation model for the magnetic memory detection of steel plate welds,and the leakage of the tangential component of the magnetic field Hp(y)in the welding process was analysed to obtain the tangential component of the magnetic signal Hp(y)and the normal component Hp(x)and the gradient value K.Then the use of normalised data processing means for different units of The residual stress and the gradient of the tangential component of the magnetic signal K are normalised using normalised data processing to investigate the relationship between the residual stress and the magnetic signal by comparison.Finally,the magnetic signals of steel plates with different lift-off heights and different weld sizes are simulated to provide a basis for testing using magnetic memory methods.This thesis uses the magnetic memory method to obtain the normal component Hp(y)and gradient K of the leakage field in two different directions,longitudinally and transversely,and then compares the tangential component Hp(y)curves of the leakage field in different paths perpendicular to the heat-affected zone to explore the variation in the welded joint and the heat-affected zone around the weld.The influence of the detection path,lifting height and weld size on the leakage signal of the specimen was analysed and the variation characteristics of the leakage signal at the spatial location near the weld were investigated: the tangential component of the leakage signal Bx,By has a maximum value at the centre of the weld and the gradient values K(Bx)and K(By)of Bx,By pass the zero point at the weld;the normal component Bz has an antisymmetrical maximum and minimum value at both ends of the weld.The normal component Bz has a symmetrical maximum-minimum at both ends of the weld,similar to an inverse sine wave,and has an over-zero at the centre of the weld,and the gradient value K(Bz)of Bz has a maximum at the centre of the weld.Increasing the width of the weld increases the crest width region and decreases the maximum value for Bx,By,and increases both the crest width region and the maximum value for Bz,K(Bz);increasing the depth of the weld increases the crest width region and increases the extreme value for Bx,By,Bz.A comparison of the experimental and simulated results gives a good agreement.At the end of the thesis,wavelet analysis techniques are applied to the processing of residual stress signals,and wavelet denoising methods are explored for the residual stress magnetic signals.Several wavelet functions are also compared for denoising,and a wavelet function db4 denoising algorithm with better denoising performance is derived.One-dimensional wavelets are used to decompose and reconstruct the magnetic signal,so as to achieve the accurate extraction of the magnetic signal.The results show that the wavelet selective reconstruction method can be used for the processing of residual stress signals. |
