Noise Filtering in PID Control | Understanding PID Control, Part 3
Вставка
- Опубліковано 6 жов 2024
- This video describes how to make an ideal PID controller more robust when controlling real systems that don’t behave like ideal linear models.
Download Code Examples to Learn How to Automatically Tune PID Controller Gainshttps: bit.ly/2HKBh12
Noise is generated by sensors and is present in every system. The derivative in an ideal PID controller amplifies high-frequency noise. Even if that noise is relatively low amplitude, the derivative will sense it and possibly amplify it enough to impact the controller. To protect against high-frequency noise impacting the system, you can modify the derivative path with a low pass filter to reduce the noise before it causes any problems.
PID Control with MATLAB and Simulink: bit.ly/2Qg57y8
PID Control Made Easy: bit.ly/2Q7Hhor
PID documentation: bit.ly/2Qi4fcE
Watch more MATLAB Tech Talks: bit.ly/2rTc8Yp
Check out more control system lectures on Brian's Channel: bit.ly/2IUlvkw
Get a free MATLAB Trial: goo.gl/ZHFb5u
Learn more about MATLAB: goo.gl/8QV7ZZ
Learn more about Simulink: goo.gl/nqnbLe
See What's new in MATLAB and Simulink: goo.gl/pgGtod
© 2018 The MathWorks, Inc. MATLAB and Simulink are registered
trademarks of The MathWorks, Inc.
See www.mathworks.com/trademarks for a list of additional trademarks. Other product or brand names maybe trademarks or registered trademarks of their respective holders.
Thanks for watching, everyone! Write your questions here and I’ll try to answer if I can. Cheers!
Can Kalman filter be used in this derivative step?
Mosman Xia maybe in the derivative path, I’ve never done that though. A Kalman filter is used to estimate a state using a model and measurements. So usually you’ll see it in the feedback path blending the sensor measurement with a prediction to create an optimal estimation of whatever you’re measuring. This can act like a low pass filter and remove that high frequency sensor noise before it even enters the PID controller.
Hello Brian, for my PID, I put a LPF in front of the PID controller to suppress high frequency due to sampling (avoid antialiasing). My thinking is that this will also suppress noise and eliminate the need for a filter in front of just D. Is this reasoning correct.
Dragan Ostojić yep that would work too. If it’s a linear LPF then you can put it before or after the D path because of the commutative property of linear systems. If you put it before the P and I path then the LPF would also affect those paths then as well. It all depends on what exactly you’re going for and how you want to implement it. If your anti-aliasing filter had the right cutoff frequency and roll off characteristics for your PID controller then you’re done.
Hey Brian, I never understood why it is possible to calculate a slope at the current time step. Isn't a Derivate in a controller acausal?
Your fan since 2014
I've immediately recognized your voice and I was like"...wait...is this Brian?!"
Thanks so much, I owe so much to you sir
Me sitting here wondering did I fail to learn at school or the school failed to teach 😂
Thanks Brian best lecturer out there
Thank you for this amazing one! The clarity of each and every concept is so simply explained by you, it just makes the topic much more interesting understanding the practical implementation.
These videos are a lifesaver. Thank you for making them.
great lecture. require minimal algebra for the explanations which is great for a non-mathematical literate like me. Amazing Brian!! thank you.
If you could release all these videos before my exam Friday next week, that’d be great!thanks!
Thank you Brain for this wonderful video.
Are you going to talk about discrete PID control in future?
Thanks.
I am from future. He did.
@@varokke137 Thanks..!!
Have a nice day..!!
Hi, Brian your videos about control theory is great, I hope you will talk about MPC controller
Thank you so much for your time Mr. Brian Douglas.
Great Lecture. Well explained and compact.
Amazing explanation! I was looking for just this kinda video the whole time! Thank you so much and keep doing such videos.
These videos are so clearly explained. Nice work.
This saved my semester
Parts 1 and 2 were great, but this made my head explode
Who are the people that dislike this video. Great teacher, great lecture
Once again, great stuff Brian...looking forward to the next video. Thank you!
Congratulations from Brazil.
Great lecture!
Hello Douglas .. have a good day
Wow you are pretty good explaning this thinks, thanks a lot
Thanks for this didactic videos,its very useful.
This dude's hand draw is insane.
So incredible! Your video is so goodd
Awesome Brain! keep rocking !!
I finally understand why we need a filter...
which microcontroller is best for Simulink models? I'm designing QuadCopter and I'm getting little bit to confuse that how to implement practically this Simulink model. Thanks in Advance. and your lectures are very awesome and easily understandable.
Thanks for teaching.
Hi sir, when we normally use a low-pass filter, we face the phase lagging. For the real-time signal, phase lag is undesirable. How can we eliminate the phase lag caused by a low-pass filter?
Hello, good question! Not all phase lag hurts a real-time system. If you look at a Bode plot for a low pass filter you'll see that there is almost no phase lag at the low frequencies (passband) and that there is 90 degrees of lag at the high frequencies (stopband). So if you're using the low pass filter to block undesirable noise, the low pass filter will only add lag to those high-frequency noise signals (not a big deal) and not add very much lag to the low-frequency signals you want.
Hey Brian Great video as usual.
I have one question though at 5:10 when you derivate y(t)= sin (Xt+phi) , its derivate should not be dy(t)/dt= X cos (Xt+phi) ?, or maybe that's why you shifted it with 90deg?
Thanks
Yep, exactly why I shifted it 90 deg :)
yeah but you have also written down that noise = y(t)=asin(W t + f0) which is wrong because it's cos. This particular aside, your videos are amazing and you explain everything very clearly@@BrianBDouglas
Because sin(90+fi) =cosfi
thanks for this video
Thaaaanks for the clear explanation !!!!
Thanks for making such a good video.. 🙂
Thank you
thank you do much for these videos, i learned a lot. may i ask what software are you using for this video presentation.
Thanks!
Hello Brian, I have a sensor with long time response in the system. Let's say that the system will detect an error and will activate the actuators in order to correct it. I am afraid that in reality the error will be corrected but the system will keep the actuators activated for a longer time than it is necessary and we will use a lot more resources than it is necessary. Is there any option to delay the controller's response? The sensor can't be changed, maybe there is another option for trolling this system?
do you have any book recommendations for PIDs?
This is really amazing!!
Thanks a lot 🌸
what if one limits the sample rate for the sensor signal, would that also cut off higher frequencies?
HI Brain Douglas could you please make some vedios about filters in control systems? I am looking forward to studying them @Brain Douglas
Hey Brian, love your videos. But could you make the background white so it's easier to see?
Anyways, great video 👌
You could try inverting the colors on your monitor to make the background white.
"Thinking makes my head hurt" I now know how Ralphie on the Simpsons feels.
Many thanks Brian, I need to implement and code PID in digital form via C++ or as they call it "Discrete PID" .
Would you please help, or just provide a link where I can start up?
Chilling late night watching the videos, learning a ton, love the blackboard animations
*Video hits **2:30**, My entire world turns incandescent hot glowing white*
JEEEZUUUSSS FAAAACK MAAH POOR RETINAS!!
Anyway, love your video's, learning a ton
nice video . thnx
Excellent!!!!!!!!!!!!
@Brian， Thank you for your great video. for 9:34, the alternative approach, how should it be implemented in a discrete structure? If SN/(S + N) is "c2d"ed, a differential structure will appear.
thanks
I didn’t know this was like whole thing.
i'm just trying to learn enough to set a control strategy for a refrigeration condenser. after all of my "actuator" hardware is instantiated in the application, it only asks for 3 settings P I D, which i assume are unitless whole numbers? does it differ across different controllers or is there some sort of standard in place?
where does the 90 degress come from in 5:10
I want the simulation file of adjusting the pid fines by the pso algorithm using the three-phase induction motor optimization method by direct torque control.
Can anyone help?
How to choose the value of N ??
Thank you Brian
How can we use PID block to calculate a signal derivative
And may you explain with examples
Hope you can explain about Feedforward control together with Feedback controller
(I need to improve an existing feedback steam pressure control by adding a feedforward control part acting on steam flow measurement)
I know you didn't understand and so you are here in the comments, watch it 2-3 time and with normal speed (1x), it will help really, i am the witness!
How would I implement a low pass filter in a digital PID controller?
With an FIR or IIR filter. Pre requisite would be discrete maths and knowledge on transfer functions, Laplace transform and z-transform. If you wan’t to be able to do it yourself without the aid of a program. But really by then you should probably be studying electronics engineering or control engineering 😛 but really this stuff is 5th semester stuff
isn't the derivative wrong
5:20
No. Derivative of sine is cosine. Cosine can also be expressed as a phase shifted sine. It's the same he's used there.
very very bombe
Hey bro
What is your degree
What about your education??
I am just curious!!
low pass filter
**Click** Noice!
This video was really confusing. Why did we start taking derivatives of frequency’s, when before we were taking derivative of the difference between actual and desired value? also when u were talking about low pass filters where u talking about hardware or software? cause it seemed like hardware but then u said we are multiplying the low pas filter by the derivative, which also made no sense just by itself, which seems like software, and then u started talking about wiring things in parallel and series which made me think hardware, but you started doing more math which made me think software.