A Data-Driven Approach to Multiresolution Analysis of Near-Field Scanning

Near-field scanning has been widely adopted as a valuable tool in diagnosing electromagnetic compatibility (EMC) and electromagnetic interference (EMI) problems. This paper proposes a multiresolution dynamic mode decomposition (MRDMD)-based method for analyzing time-varying near-field radiation. MRDMD executes the traditional DMD method recursively and hierarchically. The distribution of DMD eigenvalues determines the slow and fast modes in a level’s decomposition, in which the slow modes are reserved, and the fast modes are utilized to generate the input data for the next level. Finally, the multiresolution time-frequency representation of the near-field radiation field is obtained. And the spatial distributions corresponding to each frequency component are also extracted. A numerical example of turning three loop antennas on and off is conducted to validate the proposed MRDMD method. The multiresolution time-frequency representation of the near-field radiation shows a good agreement with the actual ones regarding the excitation duration and frequency components. Besides, the spatial distributions in the scanning plane radiated by each loop antenna are derived correctly. Hence, our work offers a multiresolution analysis tool for near-field scanning, especially when the spatial-temporal information of each frequency component is needed.