In this paper, a novel dynamic inspection method with the help of the scanning electromagnetic thermography (SET) and an advanced data reconstruction algorithm is proposed. The new inspection method is utilized to detect the multiply motor winding defects automatically and visually. The SET-based experiment system is conducted to simulate the industrial assembly line and realizes the constant acquisition of thermograms. However, the acquired thermal images are misaligned in time and space which cannot be used for further analysis. Therefore, a new 3D data reconstruction algorithm is proposed to achieve accurate spatial-temporal calibration of the image sequence. The classical contrast enhancement methods, including Fast Fourier transform (FFT), Principal Component Analysis (PCA), and Non-negative Matrix Factorization (NMF) are employed to process the reconstructed data. The experimental results show that the proposed dynamic inspection method could provide a rapid and reliable means for the research, manufacturing, testing, and maintenance of motors.

Dynamic inspection, data reconstruction, motor winding defects, scanning electromagnetic thermography.