中文摘要：

感应热成像检测通过感应加热对试样进行激励并根据热像仪获取的温度空间分布和时间变化信息进行缺陷检测.系统中励磁线圈通常采用水冷铜管绕制,所产生的激励电磁场空间分布不均匀,容易掩盖缺陷引起的温度分布特征以致无法判断缺陷是否存在.采用有缺陷和无缺陷时的温升极差比值评价系统信噪比,通过数值计算定量分析了单根铜管和2根平行铜管引起的激励不均匀程度以及几种典型实验条件下的裂纹检测信噪比.设计了基于无缺陷场景主成分重构的激励不均匀分离算法,12种工况下的24次实验表明该算法将系统信噪比均值由0.509 dB提高到2.696 dB.

英文摘要：

Induction thermography conducts defect detection by firstly stimulating the samples with induction heating and then evaluating the abnormality of space distribution and time variation of the temperature captured with the infrared thermal imager. The magnetic field created by the water-cooled copper coil is usually non-uniform, which will influence the temperature distribution characteristics and make the defect detection fail. The ratio of the temperature increase between the defective samples and good ones is adopted to evaluate the signal noise ratio （SNR） of the system. The stimulation non-uniformities caused by a single coil and two parallel coils were investigated quantitatively with numerical calculation and the SNRs of crack detection under several typical test conditions were analyzed. A separating algorithm for the stimulation non-uniformities was designed through reconstructing the thermal image based on the principle components of the data from good samples. 24 tests under 12 different conditions show that the average SNR is improved from 0. 509 dB to 2. 696 dB using the proposed algorithm.