Sign in to follow this  
Followers 0
Macgyver BR

Plc's Fuzzic Logic

28 posts in this topic

Which are the PLC's that support Fuzzy logic? That they possess a set of Fuzzy rules Happy day

Share this post


Link to post
Share on other sites
I've never heard of it that way, fuzzy logic. Do you mean a PID loop control? PID loop control allows you to control and output, proportional to a certain given rules of input. Like a thermostat. Is that what you are looking for when you say fuzzy logic?

Share this post


Link to post
Share on other sites
Yes, exist. It has PLC that it possesss set of Fuzzy instructions. Fuzzy = misty Lógic

Share this post


Link to post
Share on other sites
Omron makes a fuzzy logic co-processor for the mid-sized PLCs (C200H, C200HS, C200H Alpha and CS1.) Model C200H-FZ001. Here is a link to the manual: C200H-FZ001 Operation Manual

Share this post


Link to post
Share on other sites
Wow...Iearned something today.

Share this post


Link to post
Share on other sites
Fuzzy Logic has a big following in Japan but has not caught on in North America. You might want to think of it as a multi-input/multi -output process controller which can also take non-traditional inputs such as time-of-day, time-to-run and total product produced. It is a rule-based system where all rules have the same chance to effect the output rather than a cascaded loop controller. I have done two projects with it: belt tensioning and chip frying. Both projects could have been done with traditional PLC control.

Share this post


Link to post
Share on other sites
New info, more PLC Fuzzy: Allen Bradley and Moeller. Reference PLC Fuzzy Moeller CPU is chip Siemens I not find name PLC's, you search send message, or post down. Moeller:The fuzzyPLC model PS4/401MM2 hardware Fuzzy Applications FUZZY APLICATIONS, SEE PROJETCS

Share this post


Link to post
Share on other sites
Delta PLC have fuzzy Logic http://www.delta.com.tw/ and it`s very useful In Temperature control . weight applications ...etc.

Share this post


Link to post
Share on other sites
Why use fuzzy logic? It is by definition 'fuzzy' when there are numerical methods of system identifaction and PID that can very precise with less effort and much greater efficiency. However, I would be interested in Jay telling us about his experience with the fuzzy logic controllers. I would like to know about tuning, software, and the results.

Share this post


Link to post
Share on other sites
Peter I know that you directed your question at Jay…but I could not pass up on the opportunity to open my big mouth…. I don't like Fuzzy logic (that I have used) I have tried using it on several occasions with Honeywell temperature controllers…with very limited success at best And now that I am ranting I also don't like acutune/autotune this is a easy out when the maintenance staff is looking for a way out (to go sit on their asses) this takes over a hour to complete it's cycle (break Time!) I have a lot of pneumatic positioners that are in a very bad environment …ok, so what it's hot, but they figure that they can postpone the inevitable (so someone will fix it for them)

Share this post


Link to post
Share on other sites
Can you go into more detail? I would like to know what limited the success. What was the definition of success? Is a fuzzy logic controller easier or harder to tune? I am a PID fan mainly because I know PID inside our and backwards. What I would like to know is if there would be benefits, real or perceived, to learning fuzzy logic? There would need to be something that fuzzy logic can do that a PID can't to make the effort worth while. All in all there are some basic principles that are the corner stone of all control strategies. I can't see what makes fuzzy so special.

Share this post


Link to post
Share on other sites
Peter, I guess it's my process more then anything else…but I don't think that it's very much out of the ordinary, I can recall several examples…they are temperature related Trying to setup air temperature controller… the air is heated by a series of oil radiators, I control the oil temp, but my header that I draw the oil from can fluctuate +/- 30deg F, so if I set up using "fuzzy logic", it will use what ever the header oil temp is at that time, so it works ok for that moment but when the header changes then the "fuzzy" is fuzzy at best Setting up a thermal oxidizer temperature controller…on this I'm trying to maintain a set point of 1400deg F, I use natural gas to achieve the set point on a cold start… after I achieve set point for a period of time then I give the enable to our process, once the process starts then I start sending voc's (acetone, MEK, CHO) to the incinerator, these VOC's are highly flammable, so a change of 2% LEL can change my gas valve 10%, Fuzzy was pathetic for the application (even Honeywell gave up rather quickly), I (we, I had Honeywell help) set up a dull loop inside of this controller, looking at the distance from set point having two PID's one inside 10deg the other out side for faster reaction, it has worked very good. I think a fuzzy app would be one that the variables don't change that often, a pizza oven or electric melting pot..., then and only then I think the fuzzy could put up a fight against your PID's I have two different types of controllers that I use KFM and Honeywell, I like them both So to answer your question " What was the definition of success?"...Failure, but keep in mind that I don't have the half of the knowledge that you do, so you may have better success then I have had. I don't know if you have been to this site but it has some info http://www.fuzzytech.com/ I got it from control.com, I have not had a lot of time to read it, it may be worthless

Share this post


Link to post
Share on other sites
I just still don't get the fuzzy logic but I have never had an application that the variables don't change constantly. Is that what is causing me not to understand or is there just no chance for me ?

Share this post


Link to post
Share on other sites
Did you see this post by Alaric? http://www.plctalk.net/qanda/showthread.php?t=23987 he has a link to a book chapter 26 is fuzzy logic, its pretty good (brief but good)...I still hold reservations on actually using it Edit: I do like the disclaimer "it is also possible to implement a fuzzy logic controller manually"...would that be setting up a PID? Edited by geniusintraining

Share this post


Link to post
Share on other sites
This topic was so interesting, I had to chime in. I first learned fuzzy logic when working on video camcorders. This was one of the first consumer applications. The way it was described to me, may help you to grasp what it does for certain applications. Assume you are panning a scene with your camera, as you know you would have several items at differing distances. Using fuzzy logic, as you pan, the camera decides (based upon rules that are set ) where the focus should be set. If you are looking at a person, but they only take up 30 percent of the screen and a wall at 20 feet behind takes up the remainder, the camera may want to focus on the wall (higher percentage) but, if the perason is more centered in the picture, then the camera will want to focus more attention there. Eventually as I panned, then the wall would get more attention and focus would change. You've no doubt seen auto-focus cameras, but the ones with fuzzy logic generally do a better job of keeping the focus on the intended target. Hope this helps some.

Share this post


Link to post
Share on other sites
Ok I guess I understand. I just don't have that many uses for it Thanks

Share this post


Link to post
Share on other sites
I see plenty of these "app notes" about how much better fuzzy logic is than PID. The problem is that they don't know how to tune a PID. Check this out: http://focus.ti.com/lit/an/slaa235/slaa235.pdf This requires a lot of CPU time to implement on a MSP430. Also, if you notice it has only an Error ( proportional ) and dError ( derivative ) terms. I can implement this in a PD controller: output = max(min(Kp*error(n)+(Kd/T)*(error(n)-error(n-1),10),-10) This simple and requires only a view multiplies and adds which is going to be much faster than all that rule checking stuff and find centroids etcs BTW, there is a lot that one can learn from the TI website by searching for application notes. Sadly this example is not a very good one. GIT, your process would not be easy no matter what method you use because the heating source changes. However, a PID has the I term which will try to compensate for the fact that the 'system gain' changes when the oil temeprature changes. To do your application properly you should have a sensor for the oil inlet and outlet temperature. Both the PID and Fuzzy logic system would perform better. GIT what you are doing in your system is adding heat at a rate proportional to difference in the inlet and outlet oil temperature times the flow. If the oil temperature drops then the flow must be increased to provide the same amount of heating. This means there must be another gain after the PID to compensate for the changes in drops in the oil. A fuzzy logic control could incorporate all the inputs into ONE fuzzy controller but keeping all the rules straight may be harder than the PID + P method I described. Finally, fuzzy logic controllers don't have a sense of history. Neither do PIDs for that matter. Compensating for dead time is always a problem. Fortunately there are ways to compensate for the dead time on the PID. See www.controlguru.com. There are more sophisticated methods too but they go way beyond just a PID or fuzzy logic control. GIT, I bet you too were fighting dead time problems. GIT, your application is as much a test of the people doing the controls as a test of which technology is best. OT, it is too bad there aren't engineering Olympics. Who cares if so and so can run a 100 meter .01 seconds than the next. Show me the person that can improve the process by .01%

Share this post


Link to post
Share on other sites
Peter, Have you seen this? http://hpsweb.honeywell.com/NR/rdonlyres/4...LoopPN05035.pdf My money would be on you

Share this post


Link to post
Share on other sites
With all the improvements Chakorules has been making I think he could come up with something like this

Share this post


Link to post
Share on other sites
with out a doubtOn my KFM controllers they have P, I, D, and "anti-reset"...i'm not to sure of the term, but it works, I'm assuming the inverse of reset?...I try and keep them moving if that makes any sense, recently we have been trying to get a closer span (depending on the recipe SP 380~480, target PV +/- 1.5deg) one that I have been working on has a setting of P~8.5, I~390, D~45 and Anti-Reset of 5, I know that the integral is on the hi side but it works, I guess thats the whole idea, get it and keep it working...

Share this post


Link to post
Share on other sites
Just thought I would throw in my 2¢ for what its worth, which isn't much. I've never come across a process application that I couldn't control with a PID. I'm not arrogant enough to say that they don't exist because there are so many areas in which I have zero experience. However I have discovered that you can be very creative with PIDs. Between schemes such as PV range dependent gain sets, cascade, low select and on occasion direct manipulation of the integral sum, you can probably cover more than 99% of the problems out there with a well tuned PID loop.

Share this post


Link to post
Share on other sites
You forgot feed fowards, Smith Predictors, observers etc. Yes, but these techniques are in addition to PID control. We recognize that PID is not and can not be the whole solution. The Fuzzy logic people only think in terms of Fuzzy Logic only. I don't see Fuzzy logic people mention Fuzzy and feedwords or Fuzzy Smith Predictors, or Fuzzy observers. Fuzzy people try to pertetuate the myth that Fuzzy logic can do everything and it is just wrong. PID can't do everything and neither can fuzzy logic. Using only fuzzy logic or PID control is like having only one type of arrow in your quiver or having a hammer as your only tool.

Share this post


Link to post
Share on other sites
Fuzzy logix designer in logix platform http://literature.rockwellautomation.com/i...br009_-en-e.pdf

Share this post


Link to post
Share on other sites
Looks good and I would invite any tool that could help me do my job, but.... I dont think they will ever be able to beat a human mind (not that they are trying), there are many variables that must come into play and can not be compensated by PV vs SP or 'good' motion. Setting up a PID loop (to me) is more by feel then reaction to the process, I use the WAG to get me in the ball park, then look at the reaction vs input to the loop.. if that makes any sense

Share this post


Link to post
Share on other sites
Comments on Honeywell: The link to the Profit Loop paper doesn't apply to the 1/4 DIN UDC single loop controllers. The paper applies to Honeywell's Experion DCS algorithms. The 1/4 DIN single loopers use an autotune, which looks at the response to a 100% on, 100% off, 100% on, 100% off cycle. The duration of the on-off cyclesis is determined by the response. The results of the autotune calc are written to the PID constant settings. Separate settings in the controller determine whether the loop is, PID, I-only with P&D disabled for SP changes, or PD with manual reset. Honeywell has a separate function called Fuzzy Logic that they explain as: Fuzzy Logic - This standard feature uses fuzzy logic to suppress process variable overshoot due to SP changes or externally induced process disturbances. It operates independently from Accutune III? tuning. It does not change the PID constants, but temporarily modifies the internal controller response to suppress overshoot. This allows more aggressive tuning to co-exist with smooth PV response. It can be enabled or disabled depending on the application or the control criteria. That isn't fuzzy logic in the generic sense, it's an applied set of rules to limit overshoot. Given that Honeywell's market is primarily a thermal market, their performance isn't all that bad when put into thermal processes, realizing that a single PID isn't going to fix what needs to be cascaded. And I recently ran into some large, slow heated wax tanks where the maximum of 50 minutes of reset was marginal. Dan

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0