A Kernel based approach for classification of electromagnetic interference signals
Ender Luzardo, Jose Luis Paredes, Jaime Ramírez · DOAJ (DOAJ: Directory of Open Access Journals) · 2007
This paper introduces Electromagnetic Interference signal classification methods for signals obtained on ribbon cables with different crosstalk configurations. The proposed method comprises two stages. The first one is a preprocessing stage that applies either Principal Components Analysis (PCA), Kernel Principal Components Analysis (KPCA) or Independent Components Analysis (ICA) to reduce the data dimension and, at the same time, to obtain the most relevant information from the raw data. The second stage, the classification one, uses Support Vector Machine (SVM) to classify the kind of electromagnetic coupling. We compare the performance of the different classification structures obtained by combining a preprocessing method with SVM, namely PCA+SVM, KPCA+SVM, ICA+SVM as well as SVM in the time domain.