Research On Planar Array Capacitance Imaging Technique For Defects Detection Of The Adhesive Layer In Composite Components

The 14 th Five-Year Plan has included the development of new materials and structures as one of the major strategic objectives,and there is an urgent demand for new composite adhesive components that are lightweight,high strength,corrosion resistant,and high temperature resistant.However,during the production,assembly,and service of composite adhesive components,the internal adhesive layer can develop defects such as holes,debonding,and cracks due to the process,environment,and mechanical impact,reducing the life and reliability of the material and even threatening people’s lives and property.The detection of defects in the adhesive layer is of great importance to ensure product quality and safety in use.Planar electrode probes with geometric advantages are used in planar array capacitive detection technology,which can be approached from a single direction and can realize large-area non-contact detection.It is a promising new nondestructive testing technology.In this paper,the scientific problems in planar array capacitive imaging are analyzed for the detection of adhesive defects in composite components,and the image reconstruction algorithm is studied to improve the quality of the reconstructed images.Further,the accuracy of the planar array capacitive detection technique is enhanced,and a technical basis for adhesive defects detection in composite components is provided.The specific research work is as follows:To address the problem that the sensitivity of the planar array capacitance is affected by the "soft field" effect,which leads to severe nonlinearity and inhomogeneous distribution,thus making the solution of the permittivity distribution difficult,a sensitivity matrix optimization method based on iterative convolutional average is proposed.Considering the complexity of the sensitivity distribution matrix calculation,finite element analysis is used to model and simulate the array sensor.The sensitivity matrix distribution is calculated,and its distribution characteristics are analyzed.The convolution operation is introduced to realize the feature extraction of the sensitivity matrix and reduce the interference of the "soft field" effect on the sensitivity matrix.Therefore,the difference in the eigenvalues of the sensitivity matrix is reduced and the uniformity of the sensitivity distribution is improved.The current convolutional average is placed back into the original sensitivity matrix to participate in the next convolution operation for protecting the continuity of the sensitivity matrix.The effectiveness of the proposed optimization method is verified through the uniformity index of the sensitivity matrix and the experimental results of defect detection.Aiming at the low image quality due to the ill-posed of the planar array capacitance inverse problem and noise interference,a TV-L1 regularization image reconstruction algorithm is proposed to solve using splitting augmented Lagrangian equivalent substitution.The TV-L1 regularization model is established using prior estimates of solution and data errors,and the ill-posed inverse problem is transformed into an optimization problem for the approximate solution of linear equations.A splitting augmented Lagrangian equivalent substitution method is proposed to solve the problem of large deviation in the approximate solution obtained by the non-differentiability of the TVL1 model.Based on the idea of splitting variables,the residual is introduced into the TVL1 model as a constraint condition.The solution space is constrained to improve the accuracy of the solution and enhance the model’s ability to resist noise disturbances.The complex constrained problem is decomposed into TV-norm and L1-norm subproblems using the alternating iterative extended Lagrangian method to improve the stability of the solution.Further,the TV-norm subproblem without a closed-form solution is equivalently substituted for the L1-norm problem with a closed solution,thus improving the solution efficiency.The proposed algorithm can not only improve solution accuracy but also enhance the anti-interference ability of the solution,achieving an improvement in the quality of reconstructed images.A fractional-order regularized image reconstruction algorithm with automatic order selection is proposed to address the problem of partial loss of data details due to equal weighting of sensitivity singular values in the integer-order regularized(Tilhonov,L1,TV,and TV-L1)image reconstruction model.The residual of the TV-L1 regularization model is measured by the seminorm weighting matrix of the Moore-Penrose pseudoinverse.The fractional order weighting of the data fidelity term is implemented,the damping of the filter function is reduced,and the weak details of defects are preserved.Considering the deviation from the desired solution caused by the empirically selected fractional order,the automatic order selection method is proposed.The fractional order TV-L1 model is solved using the splitting augmented Lagrangian equivalent substitution algorithm.The effectiveness of the proposed algorithm is verified experimentally,and extended experiments are carried out to verify the applicability of the proposed algorithm to different electrodes and different materials.A fusion imaging optimization algorithm based on multi-objective threshold planning is proposed to address the issues of data redundancy and noise enhancement caused by increased data in array electrode movement detection.For multiple reconstructed images obtained by fractional order TV-L1 regularization during the movement process,a wavelet fusion algorithm based on energy contrast is proposed to obtain the fusion image with redundant information removed.To solve the problem of increased interference and severe artifacts in fused images,a multi-objective threshold planning method is proposed to optimize the fused image by solving the optimal threshold.Considering the problems of high computational complexity and low efficiency in the process of threshold finding,the MOEA/D algorithm based on a dual operator strategy is proposed to solve the thresholds.The proposed fusion imaging optimization algorithm can achieve defect visualization of large-area samples through experimental verification.