Adaptive classification algorithm for EMC-compliance testing of electronic devices

A novel technique that facilitates near-field (NF) scanning for electromagnetic compatibility-compliance testing is described. It performs measurements in a sequential way with the aim of discovering multiple, possibly disjoint regions where the amplitudes of an NF component belong to certain output ranges. The measured data samples are used to train a classification model where each NF range is represented by a given class (e.g. low/medium/high NF amplitudes). The outcome of the algorithm is a visual map that clearly characterises and pinpoints the exact location and boundaries of each class. Such maps are useful, for example, to detect hotspots or regions that are prone to electromagnetic compatibility issues. The technique has been validated on a measured microstrip bend discontinuity.