How Does Detection Work?

I don’t see any posts from 2024. I thought maybe device detection may have improved but I guess not. I’m quite surprised at how poorly the Sence preforms. Say! It found my coffee pot. Heat pumps, refrigerator, water heater and stove are so easy to spot due to electricity use and duration. How hard can it be? I found them. Why not let me pick and Id them for Sense?

Summarizing the enhancements I have seen over the past year, I categorize as follows:

Type of Improvement Orange/Green Monitor New Markets - Utilities
New User-Visible Features Watt Check US Utility Meter Setup and UI in App
Hints of new detection techniques
UI / Palette Changes
Non-User-Visible Improvements Power Quality reliability improvements International Requirements for utilities - 240V, 3-phase, 50Hz, etc.
Scaling Improvements Scaling Improvements

Pick and ID are harder than you think, partially due to Sense fundamentals (focus on short on/off transitions and not so much on duration), but also due to noise, variability and the similarities of many of the usage signals in your house. I did the blog linked below to help users understand what to expect, plus also talk intelligently, rather than grousing, about why Sense might not be finding something in their house. One of the biggest user / Sense mismatches is that the main Sense is tuned to see short on/off details. Our eyes and brains perceive longer 2 dimensions patterns, Sense looks at multi-dimensional details of short on/of transitions.

But who knows what’s next.

Don’t believe there has been any big detection improvements this decade.
There was supposed to be a MAJOR update in 2021 but that never came to fruition.

Why doesn’t Sense let us point out the obvious devices? I have a tankless water heater. Electric. It’s never going to detect it.

Put simply, the main Sense AI detection mechanism only “sees” devices that have a narrow, 1 second or so on and off transition. So even though ons and offs might be obvious you, Sense might not be seeing them. This can be especially infuriating for EV users that can see huge usage charging ramps, sometimes over 10kW. That’s why Sense has added a special EV-specific multi-second ramp detector for some of the most common EV models.

And even if Sense “sees” the transitions, if the transitions aren’t matched (on the same as off), unique (substantially different from the pool of transitions it can see) or frequent enough, the Sense AI will be challenged to convert the transitions it does see, into detections.

If you want more answers, you’ll need to look at your water heater from the Sense monitor’s Point of View. Here’s how:

The Power Meter within the Sense phone/tablet app, shown below, is the most similar view to what Sense’s data models “see”. In the app, Sense tags transitions that likely fall into the “fast enough” category. Pairs like the +1,348W on and the -1,358W off are distinctive enough to match up and deem a detection once Sense has seen enough of them. But the dozens of smaller ones between 10W and 30W, may or may not be distinctive enough. And if some portion of those 10W-30W transitions run continuously, day and night, they can even create noise that prevents Sense from detecting other devices in that range.If a device’s ramp is too slow to get tagged, or gets tagged as a bunch of separate segments, then Sense is unlikely to detect it. The Power Meter, with its tags, is extremely useful for gaining insights into how Sense AI “sees” your home, though it doesn’t reflect exactly what Sense “sees”.

I have a tankless water heater. The on and off at the amount of watts used should be obvious. Clearly seen as “very few items in a home work like that at that level.
The washer, almost always used before the dryer. If Sense sees the dryer it should be easily finding the washer.
All our lights are new led. I don’t expect Sense to find them at all.
For the life of me it would be so much quicker to let us tag them.
Last month our total electric home bill was $78. Maybe that is giving Sense a problem detecting appliances.

@drhz2222 , plenty of reasons Sense might not be able to “see” the on/off of a electric tankless hot water heater. Some have multiple heating elements, some user PWM which means they don’t have clear on/off transitions - many variable levels. You haven’t shared any screenshots from the phone/tablet app that give clues to potential detection issues.

As for washer / dryer sequencing, you’re clearly missing a things or two. First off, Sense works with an instant detection model today , where the bubble shows up as soon as the device turns on and disappears when it turns off. That’s not possible if you use the dryer turning on as a clue that the washer was on earlier. Second, washing machines have very messy on off patterns, that absolutely don’t fit the current Sense AI model. Here’s an example of mine, courtesy of a smart plug.

You are suggesting something smart though - there might be ways to do deferred detection based on post-processing using intelligent clues. We’ll see if that happens, but it will look vastly different than the current “bubble” approach.

Why not let us tag the meter page? Turn on the washer. Go to the meter page. Tap the graph and up pops a box to inter “washer” enter. Or. Tap here for start and there for stop. Enter washer.
I chose Sense because it looked “smarter” than the other brands. I didn’t want 30 clips and wires in my panel either. Waiting for months for detection seems ……

Take a look at my smartplug-measured Washer waveform for two back-to-back cycle above. What feature would someone tap on, when that mess is all mixed in with other waveforms I the Power Meter ? The big picture is that Sense can generally “see” and detect devices that meet the rules I described, though there can be additional issues that I also describe in the article I referenced earlier.

But there are other devices that require measurement via smartplugs or other integrations today. Maybe we see new flavors of detections in the future.

Sense has added the Watt Check feature, but I put that in the Kill-A-Watt capability category:

1 Like