Home Monitoring, Without All The Sensors

Smart homes come with a lot of perks, not least among which is the ability to monitor the goings-on in your home, track them, and make trends. Each piece of monitoring equipment, such as sensors or cameras, is another set of wires that needs to be run and another “thing” that needs to be maintained on your system. There are sometimes clever ways of avoiding sensors, though, while still retaining the usefulness of having them.

In this build, [squix78] uses existing sensors for electricity metering that he already had in order to alert him when his oven is pre-heated. The sensor is a Shelly 3EM, and the way that it interfaces with his home automation is by realizing that his electric oven will stop delivering electricity to the heating elements once it has reached the desired temperature. He is able to monitor the sudden dramatic decrease in electricity demand at his house with the home controller, and use that decrease to alert him to the fact that his oven is ready without having to install something extra like a temperature sensor.

While this particular sensor may only be available in some parts of Europe, we presume the idea would hold out across many different sensors and even other devices. Even a small machine learning device should be able to tell what loads are coming on at what times, and then be programmed to perform functions based on that data.

14 thoughts on “Home Monitoring, Without All The Sensors

  1. What’s really creepy is when you realize that utility companies (and any government that subpoenas them, or anyone with enough cash) has access to this same type of measurement. Afik it also works with water; aka you can tell if someone is showering, flushing, and so on. Over a period of time, they could collect enough data to put together a profile of what you do, how often, and when.

    Oven not working? Send them adds for a new one.
    Shower stop working? Send adds for local plumbers.

    Luckily mechanical meters don’t allow for this. Certain types of smart meters might though.

    1. AFAIK none of those products that claimed the ability to deconstruct individual device data from whole house aggregate data ever shipped. It worked in a lab, but not the real world. The press announcements made it sound like they could work miracles but they could not produce useful results in a real world setting.

      1. Energy disaggregation is still an ongoing research area, so I wouldn’t count it out just yet. The Distributed and Event-Based Systems (DEBS) conference has set their grand challenge on non-intrusive load monitoring this year. There are reference datasets like NILMTLK and IDEAL that are datasets recorded from real world homes, so algorithms/ML can be tested on real world data, not lab created data.

    2. If it saves even 0.1KWh a year of energy, they can spy on me all they want. The right to privacy is a legal issue. Some people do have valid reasons to hide things (including philosophical reasons), but the rest of us should be able to use the latest data.

      I kind of wish there was an open source data analytics company that let you download an app that tracks you. I bet there’s some pretty cool stuff they could do with a few thousand people, no legal restrictions (Because they wouldn’t have to pretend they aren’t spying on you like Google tries to), and an open source for the public good kind of motive.

      Do people move around more when it’s quiet or noisy? Can we detect depression symptoms by GPS tracks? Are there some places that increase these feelings after people go there, other than the usual suspects?

  2. There are a few issues with this:
    1) some ovens will cycle the elements because a full on burn just wastes energy and is less efficient, I hope he thought about this when he coded it.
    2) when the pre-heat cycle is done, it doesn’t necessarily mean your oven is ready for cooking. Having the air inside them at temperature is not the same as being ready if the walls are not at temperature. opening the door at this stage will just mean you loose the heat and the cooler walls won’t be able to keep the oven temperature stable.
    when the walls are at temperature is when you can start cooking. loosing all that heat can really mess up what you put in there, especially if it’s a lower massed baked good that needs less than 20 minutes.

    My advice: preheat for 30-45 minutes minimum, set a timer, Alexa can do that too.

    1. Cycling the elements might actually be a sign of a PID or PI controler…once you have an MCU in the control box and it can click a relay, there’s no excuse to not use one.

  3. I use an Eastron SDM630-Modbus for this and it works great. Using RS485 it delivers a point with about 30 metrics every second. I can even pick out the freezers compressor cycle, aggregate the consumption, and warn should the compressor fail :)

    1. I have a cheap wireless temperature sensor in my fridge and pick up the data with an RFXcom or RFlink. The data is feeded into domoticz, but of course anything will do. It very nicely shows the temperature fluctuation from cooling down and waiting until the temp is above the threshold to start the compressor again. As an added bonus, i added a temperature alert that sends me a telegram message when the temp rises above 10 degrees, for instance when the door is not properly closed, or mains of compressor failure.
      Since the universe is Murphy compliant, this is probably going to happen when i am on vacation on a faraway destination ;)

  4. For me, NILM has huge potentisl for “elder care”/monitoring. If I want to know my Mum is OK, I just need to see that she has put the kettle on in the morning for a cup of tea. If she hasnt, notify me and I’ll check up on her! This for me is a potentially huge area.

Leave a Reply to aztraph Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.