Friday 15 March 2013

controller - tuning pid in systems with delay -



controller - tuning pid in systems with delay -

i need tune pi(d) gains in scheme has quite big delay. it's mutual temperature controller, temperature probe far away heater. farther info:

the response of probe delayed 10 seconds alter on heater

the temperature sampled @ 1 hz, resolution of 0.01 °c

the heater controller in pwm period of 1 hz, 10-bit pwm

the goal maintain oscillation below ±0.05 °c

currently i'm using controller pi. can't avoid oscillations. higher gain, smaller , faster oscillations. still high (about ±0.15 °c). reducing p , gains leads long , deep oscillations.

i think due delay. settling time not problem, may take time needs.

i'm puzzling on how scheme work. let's think utilize i. when probe reaches target value , output starts decrease, temperature rising other time. cannot utilize derivative term because variations slow , derror close 0 (if set dgain huge value there much noise).

any idea?

normally big delays have 2 options: lower gains of scheme or, if have model of plant controlling, utilize smith predictior.

i start modelling scheme (using open-loop steps in input) quantify delay , time constant of plant, check if sampling of temperature , pwm rate ok.

notice if pwm frequency little in comparing plant dynamics, will have sustained oscillations because of slow pwm. can check using constant input pwm (with no controllers, open loop).

edit: didn't see problem solved, i'll leave here reference.

controller delay pid temperature

No comments:

Post a Comment