I’m surprised this suggestion never gained more traction. I’d also like to be able to see the power meter minus some or all of the “known” devices. This would be helpful for looking at the unknown devices without the noise from those that have already been learned. The learned devices tend to be those with the largest power demands and/or very characterizable on/off behavior, which as a result makes them relatively “noisy” on the power meter plot.
As an example, assume I want to use the power meter to reduce unexpected/irregular power usage or to lower power usage in general. When looking at the combined power meter plot it’s dominated by the cycling of the A/C compressor and clothes dryer usage - not much to be done about these unless we want to be uncomfortable or have wet clothes. It would be nice if we could subtract out these necessity devices from the total power curve to make spotting other potentially non-essential devices that draw more power than expected or are unknowingly turning on when not in use.
I first had this desire after seeing a few odd voltage dips and spikes (~ ±60-80 W) over a 1-2 hr period of laying in bed watching TV. Given that I hadn’t turned anything on/off (or even moved for that matter), I got curious as to what these could be. I tried looking for a pattern of this behavior over a larger time scale but quickly realized that identifying such patterns wasn’t going to be possible while overlayed with larger devices like my A/C.
A nice way to implement this may be to add a ‘Filter’ button to the power meter that would allow the user to select which devices are being shown in the power curve. Taking this approach would also incorporate the benefits mentioned in the previous suggestion below, which would also be helpful in solving the problem I encountered with trying to ID the individual source of the behavior I was seeing. For example, we could then uncheck all known devices other than ‘Refrigerator’ and see if the pattern of behavior from an unknown source correlates with the refrigerator running. If it did, we may be able to conclude that the unknown source is a defrost cycle, evaporator fan, etc.