I’m going to chime in one more time… Tell me if I’m talking too much.
It’s best if you think about Sense in terms of things it will naturally find, special cases, and things that it will be unlikely to detect in the medium term.
- Sense identifies devices based on finding recognizable on and off “signatures”, where the signature includes things like power level, phase, speed of transition, and a bunch of other features that you can’t even see in the Power Meter.
- Many devices types like heaters/ovens, AC motors/compressors, transformer-based microwave ovens, incandescent and fluorescent lights, have distinctive signatures linked to their physics. I’m a little surprised that Sense hasn’t found some of the heating-oriented things in your house, but your home may have more “noise” than most.
- Some devices like heat pumps and EV chargers have transitions that are longer than the time window of the main detection mechanism Sense uses. Sense has special “detectors” for those two flavors of devices, but those detections are more challenging since they aren’t based on fundamental physics, but instead based on electronic controllers, resulting in waveform cycles that can vary widely (heat pumps) and change with software updates (EVs). Read the Sense blogs on AC and EVs for more info on these kinds of waveforms.
- Most electronic devices that use DC (direct current) on the inside (computers, TVs, stereos, dc motors, dc-inverter-microwave ovens) don’t produce discernible physics-based signatures, so they also won’t likely be detectable by Sense.
Sense’s smartplug integrations with Kasa and Wemo are useful for watching these devices. - A single device that has multiple modes/power settings, like an electric range or oven will typically show up as multiple devices. Sense may eventually combine.
- One device might have multiple components that are detected at different times. Examples include dryers (heating element separate from the motor), fridge (ice maker heater and motor, interior light, separate from compressor(s)). Sense may eventually combine these if they are linked in time.
- Two or more devices that have a similar type and power usage can be conflated or seen as one device by Sense. All three of my garage door openers (all the same make/model) are detected as one. One of my floor heaters is detected as my dryer - both have similar 6000W cycles.
As for improvements in the data science and machine learning, you won’t find those in announcements on the forum covering new releases of the Sense app, but you can see some of the progress in the technical blogs: