研究目的
To compare principal component analysis (PCA) and independent component analysis (ICA) for radiated emissions from printed circuits, focusing on the differences in the features of the components extracted by both methods.
研究成果
The study demonstrates that for radiated emissions from printed circuit boards, differences exist between principal components (PCs) and independent components (ICs), with ICs typically exhibiting simpler spatial structure than PCs. The consistency between both techniques is achieved when the wideband Cartesian fields follow Gaussian statistics closely.
研究不足
The study focuses on a simple circuit to clearly demonstrate the distinction between extracted components. For complex circuits, these differences are harder to discern because of the complexity of the emission patterns themselves. The implementation of ICA specifically excludes PCA as a preconditioning step, which decreases computational efficiency.
1：Experimental Design and Method Selection:
The study compares PCA and ICA for analyzing radiated emissions from printed circuit boards (PCBs). The input data are measured wideband complex-valued magnetic radiated and evanescent fields with quasi-Gaussian spatial distributions.
2：Sample Selection and Data Sources:
Near-field EM emissions from a realistic microstrip inverter circuit designed for operation at 100 MHz were measured using a scanning magnetic field (loop) probe.
3：List of Experimental Equipment and Materials:
A scanning magnetic field (loop) probe was used to measure the emissions.
4：Experimental Procedures and Operational Workflow:
A planar uniform grid of 52 × 35 points was scanned in a parallel plane at 10 mm above the PCB. Measurements of the sampled complex Hx and Hy components were performed across the 600–996 MHz frequency band in steps of 6 MHz.
5：Data Analysis Methods:
PCA and ICA were applied to the measured data to extract and compare the components.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
