@dickensbill - I think thats a great idea to repurpose the solar sensor for measuring something else. I wonder how feasible it is. I might have to just try and see what happens.
Every minute take all of your data for that minute and best fit a line. Calculate the slope of that line. Record that slope for that minute. This reduces detection over an hour time frame to just 60 data points. Combined with load levels this should make detection pretty simple.
Sorry, meant my previous post to be a reply to you. lol
My take is that Sense team knows how to analyze and train against hundreds of features like the one you describe. But they had initially focused on instant on and off identification/detection, using millisecond long features. Now they are expanding to do after the fact identification using longer timeframe features. Seems like an intelligent progression, from millisecond signatures to hour long signatures.
I really like the idea of allowing the solar CTs to be used as a monitor for a single large/important load such as an EV. Especially since we’re willing to pay for the high margin additional hardware!
Yea, like my constant pressure (highly variable/proportional power) geothermal deep well that Sense can’t seem to discover/track. My solar system has its own excellent monitoring anyway.
Since Sense isn’t trying to break down the solar capture into components anyway, it would seem that this should be a simple software change….just make it a namable load rather than a fixed source.
What about people with both solar and an EV?? I’d guess that’s not an insignificant fraction of users with solar or EVs.
A third set of ports?
One could argue that Sense could include more than two current monitors, but the device hardware only supports two now. So, that would be a different device, not just a simple programming change.
Best of all would be Sense getting enough smarter so that it would work the way their vision (and marketing) call for it to. After a year, I’m not that optimistic, but I do see some recent customers having some success…hmmm.
I have roof solar monitored by Sense, and a Chevy Bolt EV, which is charged with a ChargePoint home charger. While I know exactly the pattern of charging of my EV using ChargePoint’s software, and I can see exactly the same pattern via the Sense meter, and I can see the same pattern using the OBDII port of the car, Sense has not discovered the car, after nine months. At one point Sense used to show you devices it was learning, but doesn’t now, and it was showing that it had detected an EV and was learning it. It seems to me that Sense AI is dependent on human intervention, server side for a lot more than I thought.
To Sense:
I have a 2018 Honda Clarity PHEV charged via a Chargepoint Home and would be more than happy to help collect data on this. I typically charge starting at 10pm weeknights and 7pm on weekends (scheduled) but may charge at other times.
I can provide data from the charge-point home and more precise timings or anything else you may need. Just let me know, I want to help you become the best there is and grow this company moving forward. I am also a Solar user. I am also willing to collect as much data as you can use just PM me and let me know. I may be a little slow in reacting but can collect and will provide manufacturer and model numbers etc off of the devices I have.
Matt: I own Chevy Bolt EV. Looks like sense is not detecting that. When will you add the capability to detect chevy bolt. It is charged using clipper creek charging station.
Thanks
Thanks for the offer! I’ll talk it over with data science and see if we can put your data to use.
@seaacs we have one of our data scientists focused purely on EVs at the moment so hopefully we’ll be getting more online soon. Unfortunately, EVs present some brand new problems but we’re working hard to solve them.
I have solar and a 2nd Gen Tesla Wall Connector. Would the connector eventually get tracked as well? Does solar tracking get in the way? (Or perhaps I’m misreading this thread and it will eventually get tracked.)
I also ask because I’ve had it for about a month and a half and I charge every evening. Haven’t seen it show up just yet.
I believe the above conversation was some theorizing on a better way for Sense to track EVs (with separate sensors like we use for solar). If you’re using your solar CTs on your solar feed, that shouldn’t present any issues detecting your Tesla. What model do you have?
3rd that! Love the Sense monitor. Have been running since February 2018. Three of my largest circuits are yet to be established though.
Our Daikin SmartSource 2-Stage Model GTH geothermal water furnace is not registering.
We have two different EVSEs for charging our Gen 1 Chevy Volt. We use the Level 1 that came with the Volt and a Myenergi zappi 32 amp Level 2 EVSE. The zappi uses CT clamps like the Sense monitor if that helps. It uses them to monitor both the household usage and variably charge with only excess solar power.
Looking forward to monitoring these heavy users.
I’ve been waiting almost 2 years for it to detect the Chevy volt. Ive learned sense doesn’t learn, it has no machine learning capability, it’s vapor ware.
I have a Tesla Model S and Model 3. No detection of these major loads. Not being able to track EV charging energy is a major issue. I suspect the the percentage of Sense users with EVs is way above average.
We really hope to see our Prius Prime plug-in hybrid appearing on our Sense, too!
Is the work you have done on the EVs so far leading toward that any time soon?
Thanks for all that you do!
I’m not sure what a “solar CT” is. I am not theorizing so much as probably being confused.
I have a Model 3.
Current Transformer. Just like Sense uses to detect the incoming current to your home.