I am slowing going into the software implementation phase of my floor heating controller project. Since my floor has very high heat capacity, a simple bang-bang controller (like bimetal thermostats) will not work well. I think I will have to implement a PID controller. I have a strong mathematical background and no issues with understanding and implementing the principle, but I wonder if there are any useful resources how to do the parameter tuning?
There's this: PID AutotuneLibrary.
Disclaimer: have not used it.
PID is best where you have proportional control of your control elements. Since your valves are set only to on or off, you're going to have to use the PID output to give you ersatz PWM with a fairly long window to avoid thrashing the valves.
I'd think that your situation will be complicated further by having multiple heating loops. PID doesn't know anything about the system it controls so you may find that the individual controllers become confused as the others call for heat and impact the temperature.
PID may be good enough for your purposes but I'd be more inclined to write a more holistic controller that does model your apartment and heating system. Also, as I think was suggested in another thread, you may want to consider outside temperatures and possibly the forecast too.
wildbill:
There's this: PID AutotuneLibrary.
Disclaimer: have not used it.
Thanks. Will look into this. Problem might be that I cannot have it arbitrarily test my system (I mean, I live there, it cannot go below 18 and above 25deg, i guess ...)
wildbill:
PID is best where you have proportional control of your control elements. Since your valves are set only to on or off, you're going to have to use the PID output to give you ersatz PWM with a fairly long window to avoid thrashing the valves.
Yes, as you read in my initial topic, I know that. I aim for 1h PWM cycle length for now.
wildbill:
I'd think that your situation will be complicated further by having multiple heating loops. PID doesn't know anything about the system it controls so you may find that the individual controllers become confused as the others call for heat and impact the temperature.
True, there might be a situation where one heat cycle takes over another room ... Actually, I did think about having ONE control loop (based on average temp, or temp of the biggest and also most cental room) and weight the rooms ... I also think I suggested taking into account outside temp myself.
Still, I need parameter estimation, and it would be good to know ho it is generally done to adjust it to my own special case.
Take a look at the Wikipedia article on PID controller - it presents a number of tuning methods.
wildbill:
Take a look at the Wikipedia article on PID controller - it presents a number of tuning methods.
The English Wikipedia is indeed quite extensive there. Thanks!
Åström–Hägglund sounds promising, since it can essentially run from the current bang-bang state.