(01.10.2017, 14:58)lek Wrote: Ahoi,
it is getting more important every day.
Temperatures changes outside and I need to adjust schedules manually. Is there a working auto learning heating script for Openrb?
Firstly in my opinion you should add trends(if you don't have it yet) Outside temperature(and also some information about other weather data like wind, humidity could be also helpful), Inside temperature, Setpoint and maybe what is the status of the valve or fancoil. Than it is possible to use this data via script and also you can check it by yourself in graph form.
What kind of heating type do you have?
What kind of control type? If you have PI or PID then it has also some inertia which should be taken to consider. Especially on start.
And also please tell what is your expected sheduler(e.g. at 8 should be 22°C in the Kitchen) and how looks the current situation(how you are changing it manually, what principle and if your change is enough or too low and at 8 you have e.g. 20.5°C).
And also what is the minimum temperature(night mode)?
I would think such could work like below:
1:00 - scheduler correction based on
yesterday outside and inside temp,
current outside and inside temp and the
weather data for today,
calculated heating start time - start
heating and save current conditions to storage,
8:00 - checking if temp is achieved or not and also save this info and current condition
if not then algorithm waits for
And on start it could be very simple(but can be helpful in your situation when you are changing temperature manually) like:
yesterday
heating start time: 7:15
achieved 23 minutes after it should so tommorrow(if conditions would be ~same) start heating 23 minutes earlier. You could also check if your setpoint is not too earlier.
But in my opinion it should be based on trend logs.
This is only some sketch.
