Research On Defect Reconstruction Method Of Eddy Current Pulse Thermography

Titanium alloys,austenitic stainless steels,and other non-ferromagnetic metal materials have been widely used in aerospace,petrochemical,nuclear industries,among others,due to their excellent corrosion resistance and high-temperature resistance.Among the common types of defects,stress fatigue cracks and corrosion defects often originate from surface cracks,gradually propagate inward,until fracture occurs,leading to catastrophic accidents.Therefore,non-destructive testing and quantitative evaluation of defects are of paramount importance.Eddy current pulsed thermography(ECPT)is a nondestructive testing method coupling electrical,magnetic,and thermal fields.It has been applied to defect detection in non-ferromagnetic metal materials due to high efficiency,high spatial resolution,noncontact,and non-pollution advantages.However,existing thermal image processing algorithms often can only identify hotspots at the endpoints of defects,making it difficult to accurately reconstruct the defect profile.The reason is that existing methods mainly rely on data features to identify defects statistically,lacking analysis of physical mechanisms and application of physical models.In this dissertation,the problem of reconstructing surface defects is analyzed from the perspective of physical mechanisms.The analysis shows that the coupling between eddy current field and defects is tighter,with less noise interference and stronger overall correlation compared to the thermal field,making it more suitable for reconstructing surface defects.Since the eddy current field is directly related to the material’s electrical impedance characteristics,drawing on the electrical impedance tomography(EIT)method,the material’s conductivity distribution is reconstructed using measured eddy current fields,thereby achieving the reconstruction of defects.The main research contents of this dissertation include:1)Acquisition of Measured Eddy Current Field:To solve the issue of blurred eddy current field caused by thermal diffusion effects,a separation algorithm for eddy current field and thermal field is proposed based on in-depth analysis of the coupling relationship between eddy current field and thermal field in eddy current pulsed thermography detection.The algorithm successfully extract the eddy current field from the temperature gradient at the beginning of heating.Experiments using three different-shaped defects demonstrate the accuracy of the proposed method in extracting the distribution of eddy current field.2)Construction of Forward Problem Model for Eddy Current Field:To tackle the complexity and computational burden of existing numerical models for eddy current field,an efficient eddy current field model is proposed.On one hand,unnecessary computations around the surrounding area are avoided by optimizing local boundary conditions and constraining the modeling range within the detection area,which reduces computational workload.On the other hand,complexity of the model is reduced by approximating the secondary magnetic vector potential induced by eddy currents as uniform distribution.Experiments with three different-shaped defects show that the proposed efficient model does not significantly compromise accuracy while significantly improving computational speed.3)Design of Defect Reconstruction Optimization Algorithm:To address the issue of significant noise interference in measured data,a defect reconstruction method based on nonlinear least squares optimization is proposed.This method constructs an objective function by minimizing the residual between measured current and predicted current,and introduces Tikhonov regularization to provide a smooth prior,reducing noise interference.Experimental results demonstrate that the proposed method is able to reconstruct material surface conductivity distribution and accurately identify defect profiles.4)Design of Multidirectional Eddy Current Information Fusion Optimization Algorithm:To address the problem of information loss and reconstruction errors caused by unidirectional excitation,a defect reconstruction method based on multidirectional eddy current information fusion is proposed.This method compensates for the information loss with orthogonal directional eddy current and fuses their information through expanded sensitivity matrices to achieve reconstruction.Experiments with multiple adjacent defects,complex-shaped defects,and defects parallel to the eddy current direction show the accuracy of the proposed method in reconstructing defect shapes.5)Design of Optimization Algorithm for Three-Dimensional Defect Reconstruction:To solve the issue of inaccessible internal eddy currents in three-dimensional defect reconstruction,a defect tomography reconstruction method based on nonlinear mapping is proposed.The detection area is modeled as a three-dimensional conductivity tensor.A nonlinear mapping from surface eddy current field to each layer of the three-dimensional conductivity tensor is established using sensitivity matrices based on the three-dimensional eddy current field model.This approach reconstructs the three-dimensional conductivity tensor from surface measured eddy current directly,avoiding the need for internal eddy current.Experiments with defects at different depths,complex-shaped defects,and natural cracks demonstrate the effectiveness of the proposed method in reconstructing threedimensional defect profiles.