基于传感器阵列和独立分量分析的远场涡流缺陷信号盲分离技术研究 Blind Separation of Remote Field Eddy Current Defect Signal from Magnetic Perturbation Based on Independent Component Analysis

针对远场涡流检测中管道磁导率不均匀严重影响缺陷信号检测的问题，本文提出一种新的基于独立分量分析的远场涡流缺陷信号盲分离技术。首先利用有限元仿真对独立分量分析在缺陷分离中的适用性进行了详细分析，证实了磁导率不均匀和缺陷信号满足独立分量分析的应用前提。提出了基于实部和虚部信号的独立分量分析缺陷分离方法，并进行了仿真和实验验证。结果表明，基于独立分量分析技术能够有效的从磁导率不均匀区域分离和识别出缺陷。 According to influence of magnetic perturbation in remote field eddy current defect detection, a novel blind separation technique based on independent component analysis was proposed. Firstly, finite element method was adopted to declare the applicability of independent component analysis in remote field eddy current. Secondly, a new method based on real part and imaginary part rather than amplitude and phase signal was proposed, and simulation and experiment was implemented to verify it. At last, a conclusion could be deduced from simulation and experiment results as followed: the influence of magnetic perturbation could be eliminated significantly by using independent component analysis.