This is not an "introduction to state space models" video. You need to first talk about what does a state mean what does state space mean, why and where this model can be used etc. before diving right into the algebra behind the model.
@@apm An example is T(da/dt)=kq+b-a with T=1, k =0.2 and Ts=0.25. I want to make it into a discrete-time model where a(k+1) = 0.7788a(k) + 0.0442 q(k) + 0.2212b(k)
Hi, how to identify the state variables in the state space models' equations ; means to know that which state of the system is being represented by the variables?
If you identify a state space model from data then the states have no physical realism. However, if you obtained the model from a physics-based model then the state corresponds to the original variable that you began with: apmonitor.com/pdc/index.php/Main/StateSpaceModel
shouldn't' both elements in the bottom row of matrix A for the second-order differential be -1/tp2 and 1/tp2? i.e. both denominators be tp2. if not can you explain
The output equation is typically a condensed identity matrix with ones for the states that are measured. In this case it is y=x2 but it could also be y1=x1 and y2=x2 if both states are measured. In that case, it would be C=[1,0 ; 0,1]
I have a multi agent system problem that models the agents with an equation similar of what you describe, I wonder if you would be able to cooperate to model this problem. Thank you.
First be a Teacher first , make sure you have enough knowledge . Here people came for clearing their problem ,not to just memorize and write equation out of nowhere . we are students not memorizing machine
Thanks for the suggestions. There is additional content here if you need more information, like the derivation of state space or more examples: apmonitor.com/pdc/index.php/Main/StateSpaceModel
Why is the function F(x,u) = 0 at "the nominal values" of x and u? What if F(x,u) = x^2 + u^2 + 1? Then F(x,u) will never equal zero, so why can we assume this to be zero? Also, you are taking a taylors approximation to make the nonlinear function F linear. But... does this assume that the derivatives of F are not linear? Rest of video is very, very well explained and clear though. Thank you.
Thanks for your question. It is zero because the nominal conditions are typically chosen at steady state conditions where the time derivatives are zero. dx/dt = 0 = F(x,u) please see this additional information on linearization: apmonitor.com/pdc/index.php/Main/ModelLinearization
This is not an "introduction to state space models" video. You need to first talk about what does a state mean what does state space mean, why and where this model can be used etc. before diving right into the algebra behind the model.
Great suggestion! I do mention more of this background information on the associated web-page: apmonitor.com/pdc/index.php/Main/StateSpaceModel
I went through your page but sadly what I was talking about does not appear there as well. A "state space model for dummies" video would be excellent.
ua-cam.com/video/p9qrHdPEe28/v-deo.html
@@valmormn I searched for a long time to find a comprehensive but understandable introduction. This video is really great, thank you very much!!
We've all been there bro, welcome to university
This is honestly a bit difficult to follow for beginners but its a lot better than the others i have watched
Thanks for the feedback. There is additional content here: apmonitor.com/pdc/index.php/Main/StateSpaceModel
Thanks for all your clear videos and resources! Greatly appreciated!
You are a life saver!!Can't thank you enough for the explanation in such a short video!!Kudos!!
I like the sound that the pen makes on the touchscreen.
That was back when I used the microphone on the tablet for the recording. Much less pen sound in the new videos.
Do you have any teaching materials from a differential algebraic equation to a discrete-time model?
Is this what you need? apmonitor.com/do/index.php/Main/OrthogonalCollocation
@@apm An example is T(da/dt)=kq+b-a with T=1, k =0.2 and Ts=0.25. I want to make it into a discrete-time model where a(k+1) = 0.7788a(k) + 0.0442 q(k) + 0.2212b(k)
Hi, how to identify the state variables in the state space models' equations ; means to know that which state of the system is being represented by the variables?
If you identify a state space model from data then the states have no physical realism. However, if you obtained the model from a physics-based model then the state corresponds to the original variable that you began with: apmonitor.com/pdc/index.php/Main/StateSpaceModel
Hi , thank you for this video
I don't know how anyone who doesn't already know state space forms would follow this. No insight, just regurgitation of the mechanics.
This is certainly a video for beginners so it probably didn't meet your needs.
Eh, if you know differential equations and matrix algebra you can follow it.
@@tag_of_frank I agree. It's introduction to state space models, not introduction to differential equations and matrix algebra.
shouldn't' both elements in the bottom row of matrix A for the second-order differential be -1/tp2 and 1/tp2? i.e. both denominators be tp2. if not can you explain
Yes, I fixed it in the solution here: apmonitor.com/pdc/index.php/Main/StateSpaceModel
@@apm - many thanks - and you're website is great - love the worked examples - very helpful
can you please explain how you get the output equation?
why does y=x_2 and not y=x_1?
The output equation is typically a condensed identity matrix with ones for the states that are measured. In this case it is y=x2 but it could also be y1=x1 and y2=x2 if both states are measured. In that case, it would be C=[1,0 ; 0,1]
@@apm how do you know which state is being measured?
I have a multi agent system problem that models the agents with an equation similar of what you describe, I wonder if you would be able to cooperate to model this problem. Thank you.
We have a new python package that helps with modeling ARX and state space. Check out the Gekko optimization Suite.
@@rrc Actually I need to simulate it in MatLab
@@ingridbernal Here is some help with MATLAB / Simulink: apmonitor.com/che436/index.php/Main/ProjectProgramming
"Introduction", clearly......
Here is additional content: apmonitor.com/pdc/index.php/Main/StateSpaceModel
First be a Teacher first , make sure you have enough knowledge . Here people came for clearing their problem ,not to just memorize and write equation out of nowhere . we are students not memorizing machine
Thanks for the suggestions. There is additional content here if you need more information, like the derivation of state space or more examples: apmonitor.com/pdc/index.php/Main/StateSpaceModel
Why is the function F(x,u) = 0 at "the nominal values" of x and u?
What if F(x,u) = x^2 + u^2 + 1? Then F(x,u) will never equal zero, so why can we assume this to be zero?
Also, you are taking a taylors approximation to make the nonlinear function F linear. But... does this assume that the derivatives of F are not linear?
Rest of video is very, very well explained and clear though. Thank you.
Thanks for your question. It is zero because the nominal conditions are typically chosen at steady state conditions where the time derivatives are zero. dx/dt = 0 = F(x,u) please see this additional information on linearization: apmonitor.com/pdc/index.php/Main/ModelLinearization
Very Helpful, thanks
thank you . very useful tutorial ..
sir can i save page from your browser
I recommend that you select the print button and then save as a PDF.
ok sir
very useful for me，thanks
sir there is an easy way to save page right click browser and save page as can i do this
Yes, you can do that. The print link formats the page without some of the extra side-bar info.
This is very good
thank you