Automatic Device Classification from Current-derived Features
Amber Frid-Jimenez in collaboration with Philip DeCamp
Figure 1: Graph of current over time for one plug (current 4). The three figures below show the classification of the devices into three types: resistive, inductive and switching. Classification was made by examining two features: phase offset between the two signals and the variability in the peak amplitude. The variability is measured as the standard deviation of peak amplitudes normalized by the range of the signal.
Figure 2: Graph showing cropped data from voltage and current sensor while the lamp was plugged in. The terminal window below shows the program classifying the lamp as a resistive device. The two numbers above the classification in the terminal window refer to the phase offset and the variability. In this case, we can see from the picture that the phase offset is 0 and the variability is 0.044.
Figure 3: Graph showing cropped data from voltage and current sensor while the fan was plugged in. The terminal window below shows the program classifying the fan as an inductive device. This is largely due to the phase offset.
Figure 4: Graph showing cropped data from voltage and current sensor while the label machine was plugged in. The terminal window below shows the program classifying the label machine as a switching device. This was largely determined by the high variability in the peak amplitudes.
