Sense detected 2 devices in 1

Device Detection
#1

Hi

I’m very new to Sense, only acquired it a few weeks ago, but I already think it’s a fantastic product/idea and am fully on board to help it learn.
In the space of about 3 weeks I got 5 natively detected devices, furnace, air fryer (Sense needs a category for this - I currently have it as a generic “Kitchen Appliance”), Microwave, Freezer and “Heat 1” (still trying to find which one is this myself).

Now for my question, I noticed the detected Freezer device in Sense is actually pulling data (turning on and logging power usage) from both my chest freezer (a Holiday LCM070LC) and my upright refrigerator (a Whirlpool WRS342FIAB00). I verified it by doing the plug/unplug test on both while Sense was detecting them running.
I have one TP-Link HS110 that I play with and am using with Sense as well, and I was wondering, would it help to plug one of the 2 appliances to it so Sense can tell them apart, and potentially re-discover the other one?
Or would it be better to delete the detected device and let it start over? Or just wait and see if it would correct itself later down the line?

#2

That is a great question and I am subscribed to see how the community replies to your query.

Mine found Heat 1 and Heat 2 and for the life of me I couldn’t think what I was using that could be a resistive heating signature. I think (and yet to verify) that perhaps they are the two heating elements in my GE hot water heater. I did monitor the graph after a shower and somewhat confirmed (in my mind at least) that one of those HEAT’s are likely the water heater.

I just ordered the additional current loop *solar sensors. I am pretty sure that I am going to install one on a leg of my water heater and the other on the three legs of my HVAC equipment (mini split inverter, Trane A/C and the Trane air handler w/resistive heat strip)

cheers!

#3

So over the weekend I went ahead and did an experiment with the TP-Link HS110 just to see the difference, and the waveforms are radically different.
Here’s the original waveform of the natively detected device

pic3
pic31242×2148 173 KB

And the next day I applied the TP-Link HS110 to the chest freezer to monitor it and see the difference.

pic1
pic11242×2148 147 KB

I still have the TP-Link plugged in but I don’t know if it is helping my Sense any, besides providing ground truth data. I guess if I unplug the TP-Link either it will start acting better (if it got smarter with the added data) or it will be back to the original waveform (if it didn’t learn yet)

#4

So this does look like a conflated device, @jorge. The compressors for both the devices are extremely similar in terms of start-up time and avg. wattage.

Personally, I would recommend adding the HS110 to the chest freezer and deleting the old natively detected device. There’s no guarantee that it won’t group these devices together in the future, but allowing Sense to reattempt detection might yield better results.

#5

Thanks @JustinAtSense. I will do exactly as you suggest.
Part of me was hoping that maybe Sense over time could adjust or self-correct the present model somehow as it receives different/conflicting data from the TP-Link.
Maybe it’s something to look forward in the future, device self-correction. (I have no idea if something like this could even be theorized, let alone put in put in practice, but it’s nice to dream)