Study On Signal Processing And Internal Defect Detection Methods For Multi-layer Conductive Structures In Pulsed Eddy Current Testing

With the progress of science and technology and the rapid development of all walks of life,non-destructive testing and assessment of key equipment and key structural components of systems are receiving more and more attention.Supported by the projects of National Natural Science Foundation of China and the requirements of practical engineering applications,the research on pulse eddy current detection signal processing and internal defect identification method for non-destructive testing of multi-layer conductive structures is carried out.Based on the principle of pulse eddy current detection technology,the influences of external disturbance and internal defect on response signal are analyzed respectively.The signal preprocessing method is studied,and signal preprocessing technique under the influence of lift off(the distance between the probe and the measured object)is emphatically studied.Aiming at the problem of qualitative detection and classification of internal defects of multi-layer conductive structures with lift-off interference,a method is proposed.The quantitative analysis method of multi-layer conductive structure internal defects under multi-interference is emphatically proposed.The PEC detection system is improved and optimized.The main research work and innovation of this paper are as follows:(1)The principle of pulsed eddy current testing technology is studied.The typical time-domain characteristics of typical pulse response signals and the influence of external disturbances and internal defects on the response signals are analyzed.The above work provides a positive guide for the pulse eddy current testing of multi-layer conductive structures.(2)In the pulse eddy current scan detection process,lift off is not static,but dynamic fluctuations,and causes the detection signal to fluctuate greatly.Therefore,in view of the dynamic change of lift-off interference,the existing lift-off elimination technology is analyzed and compared.Based on this,combined with the positive analysis of lift off in the previous section,a lift-off prediction method based on partial least squares is proposed.According to the response signal under the unknown lift off,the lift-off prediction model can be used to predict lift off.The method is mainly applied to suppress or even eliminate the influence of lift off on the pulse eddy current test results in the case of unknown or uncontrollable lift off in the actual test which lays the foundation for subsequent lift-off elimination and defect detection.(3)On the basis of the prediction of lift-off height,the qualitative detection and classification of internal defects in the multi-layer conductive structure with lift-off noise are further studied.Through the analysis of the extraction effects of different types of feature in the response signal,try to establish the mapping relationship between the feature of the response signal and the internal defects.The method of qualitative detection and classification of internal defects based on combined feature is proposed.The experimental results show that the method can effectively distinguish qualified products and unqualified products in the thermostat when the lift-off height is unknown,and effectively distinguish the different types of defects in the unqualified products.(4)In order to solve the problem of quantitative analysis of internal defects in multi-layer conductive structures under multi-disturbance,the detection environment in which many kinds of interference in the process of pulse eddy current testing is introduced,and the limitations of the traditional quantitative testing methods are also analyzed.Taking the three-layer conductive structure with through-hole as an example,the influence of through-hole structure in the test sample on the response signal is taken as an interference.Considering the existence of lift-off interference and through-hole interference,based on the traditional methods and the advantages of the algorithms in the first two parts,a quantitative analysis method based on multi-probe is proposed.Among them,the support vector machine regression algorithm based on particle swarm optimization is used to construct the quantitative model between the combined feature and the internal defect length.The problem of quantitative detection of the internal defect length in the multilayer conductive structure is solved under the condition that the lift-off interference and the through-hole interference coexist.The validity of the algorithm is verified by experiments.(5)In order to improve the stability and portability of the testing equipment,the existing system of pulse eddy current testing is improved and optimized on the basis of the laboratory prototype system.The hardware circuit of lower computer detection system has been optimized design and integrated development.In summary,through the lift-off prediction,the combination of feature extraction and multi-probe detection ideas,the thesis study the multi-layer conductive structure of the pulse eddy current signal processing and internal defect analysis method to effectively achieve the internal defects qualitative detection,classification and quantitative analysis.Moreover,it enhances the detection ability of pulse eddy current testing technology,and provides technical support for further testing applications in the future.