12/10/2010 9:02 AM | |
Joined: 4/24/2008 Last visit: 5/16/2024 Posts: 209 Rating: (16) |
it looks like Chien, Hrones und Reswick. There you need the point of inflection for calculate the PID parameters. |
|
|
12/11/2010 5:24 AM | |
Joined: 1/28/2009 Last visit: 5/20/2024 Posts: 6840 Rating: (1359) |
Dear member Check the attachment for a assumed process reaction curve and tangent line in several point of the curve.Inflection point has maximum slope of tangent line regards to other points.To calculate the slope of tangent line in every iteration of PHASE 2, difference of new and old values of process devided by CYCLE or time between two iterations.For each iteration ,we calculate a slope and compare it with previously updated max slope.This part of algorith is an old trick "How calculate maximum value in a group of numbers". Maximum slope of tangent line occur in inflection line and iteration after that, this slope is less than max value and inflection line verified and phase3 started. Hope this help Best regards Attachmentinflection.zip (240 Downloads) |
12/11/2010 8:05 AM | |
Joined: 1/28/2009 Last visit: 5/20/2024 Posts: 6840 Rating: (1359) |
Dear gandalftaurus
Best regards |
1/2/2011 12:50 AM | |
Posts: 3 Rating: (0) |
Hello again! I'm very thankful for your explanation but still I have a few unsolved problems. Firsty, how FB58 deals with noises. During few experiments I was trying to calculate the slope by myself and compere with controllers results but there was a serious defferences. My algorithm to calculate the slope of tangent was as following: (actual_measured_value - old_measured_values) / time_between_this_two_measurement In the atachment I enclose my result of measurement the PV value, slope of tangent calculated with algorithm which I mentioned and also the part of data block where you can see value of parameters calculated by controller. Why controller said in the 18'th second the slope of tangent was the highest? It seems to me that, PV or rate of change of PV is filtered in same way but how? The second doubt is, how controller estimates delay time and time constant knowing time when point of inflection occured? Finally, the last question: in the manual time constant is marked as TA, but in the comments column of data block the TA is descriped as recovery time. I fell very confused because I always thounght that recovery time is the time needed to setting up new steady state after step change. Is my thinking wrong? I hope you will help me to deal with my doubts as you did it before. Best regards, gandalftaurus Attachmentslope_results.zip (154 Downloads) |
1/2/2011 9:17 AM | |
Joined: 1/28/2009 Last visit: 5/20/2024 Posts: 6840 Rating: (1359) |
Dear member For a detailed and academic material on this topic , check Towards ‘plug and control’: self-tuning temperature controller for PLC By Bernd-Markus Pfeiffer in international Journal of Adaptive Control and Signal Processing Volume 14, Issue 5, pages 519–532, August 2000. Also check: EXPERIMENTAL EVALUATION OF A PLUG&CONTROL STRATEGY FOR LEVEL CONTROL. about FB58 and its parameter, manual is the best reference. If I get more information, I will let you know. Regards |
Follow us on