- cross-posted to:
- aboringdystopia@lemmy.world
- cross-posted to:
- aboringdystopia@lemmy.world
Software developer and open source proponent Jeff Geerling recently purchased a new Bosch 500 dishwasher, only to find it required an app to access certain features. This is his story.
It is based on a machine learning task called classification.
The reason that a machine can detect a face or a cat in a picture without seeing every cat.
Power meters can measure energy usage at high frequency, this gives it access to a lot more data to train on.
There’s a lot more raw data present in a couple of pictures of a cat than what a power meter has access to, however.
The meter can only see overall amperage draw, and without something to reference that against, it’s hard to know what’s using all the power.
Was that the dishwasher cutting on, or a chandelier with 20 incandescent bulbs? A microwave, or a hair dryer? Air compressor? Battery charger? Vacuum cleaner?
There are lots of options for things that use power, and any inferences you could draw off of power usage makes too many assumptions. For instance, power draw is increased by the amount of conductor between the thing drawing power, and the meter. So a hair dryer can draw more amps when used in an outlet farther from the meter vs if it’s connected to an outlet right next to it. Plus, things draw more or less power based on the work being done. A drill spinning freely will draw less amps than a drill actively drilling into something.
There’s just too many variables. The best you could hope to achieve is have a computer say “this household’s power draw at this time could have been this selection of different combinations of power draws” which isn’t very useful, especially considering how efficient things have gotten. How is the meter to know the difference between me turning on my outdoor lights (4x120w bulbs) and my computer running at full tilt (my high end GPU and CPU consume almost 500w at full load)?
https://www.sciencedirect.com/science/article/abs/pii/S2352467719300748
TL;DR: Math
TL;DR, it’s not nearly as granular as you suggest:
https://ars.els-cdn.com/content/image/1-s2.0-S2352467719300748-fx1_lrg.jpg
They can generally characterize the probability that the load is for certain things, but they can’t say that your power consumption is because you’re using a vacuum cleaner and 7 LED bulbs. They estimate the percentage of your overall consumption that is used by certain things. It’s not the same as feeding a LLM a few cat pictures and getting it to identify a cat.
That paper is specifically about low frequency data (under 1 hz) so it does not include fast transient events. Because of that, the amount of information that you can learn is limited.
High frequency sampling can capture fast transients, startup transients and information about the circuit harmonics. This provides a lot more data points to extrapolate from. Modern smart meters are available with high frequency (several kilohertz) measurements, they may not be deployed in your utility section, but they are used.
There is more to NILM (https://en.m.wikipedia.org/wiki/Nonintrusive_load_monitoring) than that one paper.
From the wiki:
My utility company told me they could do this, but I know for a fact they cannot. My power meter broadcasts its instantaneous reading in short plain text packets at a frequency once every few seconds. They told me all my power usage was hot water. I’m sure it’s HVAC and computers, which didn’t even show up in their list.
It depends on the kind of meter that your utility uses and how they have it configured.
Smart meters can measure that data at higher frequencies (up to several kilohertz) and from that data you can detect signatures that identify devices inside your house and even, in some cases, what they’re doing. For example, when your washer turns on it runs a pump (which draws a specific load) for a set amount of time and then goes through a cycle of running a motor to agitate the load, which draws energy in a specific way as it turns back and forth. When you turn on an LED light, it runs at a steady rate and draws the same amount of energy. When your AC runs it draws a different amount of current than the washer or LED.
With enough data, over time, you can determine which devices are in the house and when compared to a database of known signatures you can classify the device. Ex: all Samsung Refrigerator Model 23e4234 work the same way so once you identify the signature of one you can identify others.
Here’s some articles talking about it:
https://www.sciencedirect.com/science/article/abs/pii/S2352467719300748
https://energyinformatics.springeropen.com/articles/10.1186/s42162-019-0096-9
Also, just because your meter is only normally reporting every few seconds it doesn’t mean that it isn’t capable of recording the data faster for diagnostic purposes (“Why is this house suddenly using 10x the power?!”) or law enforcement purposes (“What house in this neighborhood is using HID lighting?”).
Not all meters have this capability. Old style meters with a disk don’t record data at all and some of the older smart meters can only sample at lower frequency. You can do the same math on the lower frequency data but if you can’t measure fast transient events you lose some of the more specific capabilities (like knowing the exact model number of your refrigerator).
You’re gonna have to redo that last sentence there. I’m not catching the drift.
What data are they gathering? Like, what specific info from the appliance can tell the power company what it is you are running?
Smart meters can measure, instantaneously, I and V at high frequencies (up to several kilohertz) and by looking at long term and transient event signatures in that data it is possible to classify the loads as coming from specific kinds of devices (down to individual model numbers for known devices).
Even if you have an older smart meter (not the analog ones) that can only sample at a few hz you can still do the same kinds of things but with less accuracy:
https://energyinformatics.springeropen.com/articles/10.1186/s42162-019-0096-9