How smart/ learning is the thermostat

Titsken

Hello, first question is how smart the thermostat is? Does he continious learn and adapt his heatingcurve or is it a one time learning for one weak or so and then used that for the rest of the time.
Why? After two weeks of use the thermostat can't get the temperature right on the disired time with early start. How can I get it right or reset the things Tado learned and start over his learning curve.

bArt

@Tado: can you please explain how the learning process is in any way related to the fact that you are using a PID-controller?

PID is referring to the way the actual regulation-loop is controlled. The learning part should automatically adapt the P, I and D parameters. In my experience the P, I and D parameters are not automatically adapting at all. The system behaves exactly the same all the time.

The early start depends on the type of thermostat you are using. For the TRV it is not working at all (confirmed by Tado; they are working on that for years now without a fix). On the normal thermostat this works kinds of OK for me.

Germán

@Titsken The thermostats’ control algorithm does not cover perfectly all possible heating setups. In some cases it may result in that our calculated behavior doesn't get the expected reaction from the heating system, as there will be an influence from other factors affecting every heat demand. These factors could include insulation, room size, ceiling height, ventilation, etc...

If you are experiencing that your system is not reacting up to the expected behavior, please contact our support so we can investigate it for you.

@bArt As a PID control calculates dynamically the required parameters for each heat demand it also influences each action from the system; but your point is right, mentioning it was not really the best answer to the original consult and that's why we've decided to redo our post to try to clarify.

matthewmorek

Does tado° learn about how quickly a given room can heat up to make sure that it only requests enough to maintain a desired temperate?

I've noticed that my bedroom when set up to 20.5°C in the morning heats up all the way from 19.5°C to 21.5°C by the end of the heat cycle, and this makes it somewhat uncomfortable at times (this is with a wireless temperature sensor and a thermostatic valve both in the same room).

It would be great is tado° could not just rely on built-in algorithms but actually learn about each room separately, making sure temperature is maintained optimally and efficiently; after all this is why we have invested into a "smart" system and why most of the computation requires "cloud" options (it might be too big for a simple IoT bridge).

waterescape

I have the same question @tado .

I only have a room thermostat with floor heating. How long before tado knows when to start the early start to get the room on the right temp? This morning it started 30m before the required temp, but with floor heating you need way longer then 30m in the morning.

So how does Tado learn and how long before operational? And does it learn / adjust itself in time (based on external factores like sun and temp outdide?

Beside that, it is set to 19.5, but is still heating while it is 20.0.

Unknown

This content has been removed.

Schippie

Hi @GrayDav4276 I read on this forum that the wavy lines actually represent the 3 steps in which the radiator valves can regulate the flow in the radiator. What you're saying is that's not the case & they correspond to the power demand. Interesting. Hopefully someone from Tado can clear this up. Can you quote the response from customer service here?

Schippie

@waterescape I stopped waiting for the algorithm to learn after a week or so. Nothing happened. Always 1,5-2 hrs late for floor heating to reach the temperature. Conversely, the radiator valves started about 3 hrs early and reached set temperature in 20 minutes. Also no noticeable change. Not using early start at all right now.

Unknown

This content has been removed.

Schippie

Thanks @GrayDav4276 . Does explain why early start for the radiator valves feels so useless. The radiator fully opens, and heats like crazy so the set temperature is reached in 20 minutes. I do wonder why the algorithm insists on always starting 3 hrs before it needs to. It's not actually a learning algorithm at all, I think.

Unknown

This content has been removed.

Schippie

True that. I've tried early start for the radiator valves for a few days, but nothing changed in the early start sequence. If an algorithm can't see that starting to heat 3 hours before a set time, then reaching the desired temperature in 20 minutes (so it's 2hrs 40 mins early) is something to learn from... then I don't know what is.

Conversely, my thermostat is connected to floor heating, and there the algorithm insists on starting about 30 minutes before the set time, only to reach the desired temperature a full 2,5 hrs late. I've had it running for a week, and literally nothing changed in the early start time.

My conclusion: there is some sort of algorithm that determines the early start time based on the actual temperature vs desired temperature, but it doesn't have a feedback loop at all.

Unknown

This content has been removed.