Derivative of block at time is Inf of NaN. Stopping Simulation. There may be a singularity in the solution. If not, try reducing the step size (either by reducing the fixed step size or by tightening the error tolerances.)
I used the fuzzy logic block in my simulation. when i delete the fuzzy logic, my simulation runs but with it is used again i received the error. what should i do?
regards
Click on that error and find the point its causing error. Add a "UNIT DELAY" block at that point or nearby and run simulation again.
This error usually occurs in unstable systems. There may be a positive feedback or something like that. You might check if your fuzzy controoler is properly tuned.
hi
maybe ,your FLC inputs structure are limited but the real inputs have values are not in range of FLC inputs design, this is the cause of NAN errors
you should be added gains to the FLC inputs to make your inputs in the range of FLC inputs structure
You may make the step size adaptive and try. Further, the following two links may be helpful to you:
https://in.mathworks.com/matlabcentral/newsreader/view_thread/241876
http://in.mathworks.com/matlabcentral/newsreader/view_thread/263947
Dear researchers thanks for helping.
somtimes my simulation runs and somtimes doesnt !! why??
That really depends on the task you are trying to achieve. Could you give an elaborate explanation of your goal? Maybe a screenshot of your model or the model itself. that would be very helpful in defining the source of the problem.
Mr. Denys, my goal is regulation of voltage and frequency of a microgrid. i used a GDC controller for that. in the GDC we need R and X of the line, in the complex sys,calculating the R and X is difficult so i used a FLC controller for optimal value of R and X. my sys without FLC runs without any problem but with FLC, somtimes runs and somtimes doesnt and i get the same error i asked at frist. i used different solvers like ode45, ode23t... i used unit delay and gains for putting inputs in the rage in FLC but i dont result. just my simulation runs sometimes without changing the model.
If I understand correctly, you are referring to the problem you have previously asked question about, i.e. using the fuzzy logic to define the gains of a droop controller. Are you using the same rulest and membership functions distribution as shown in the article you were referring to? If so, there may be a problem there. The variable-step solvers you were using may sometimes come to the situation, where you have a division by zero in the controller (when X=0). Firstly, I find it odd, that there are only positive values of fuzzy controller inputs. Perhaps, the absolute values are used? Secondly, the range for estimated X and R starts at 0 ohms. A small positive value instead of zero could solve the problem. Also try using fixed-step solvers.
If this doesn't help, could I ask you to send me your model via e-mail [email protected]?
Mr Denys, i faced with another problem. would u help me? i sent my simulation in ur mail and explained my problem. plz check it
regards
Mr. Denys
what i understood from ur explanation. i should use abs for the input values( 4 inputs delt-v, delta-f, delta-p and delta-q) and also change the range in the membership Funs from zero to a small positive value. yes?
The input signal to your block which is giving error because of getting infinity or Nan due to division be zero case. Please add a filter or saturation before it. It may work.
NaN (not an umber) is due to division by zero. Check the denominator variables in your simulation code. Check you equation carefully. If your working on simulink 2012, or 2013. Send me the code, I may help you in this regard.
Thanks, dear researchers. but unfortunately my problem is not solved.
Muhammad shafiq, im working on simulink 2014.
Mr, Muhammad thanks for caring. my problem is the previous problem that at time 0.0 is not finite. could you help me?
There is devision by zero in your code. You have to remove resolve this issue. without looking to the code I will be difficult. Btw, have you set the sampling period in your simulation?
I know my problem where it is. i used the fuzzy logic controller to calculate R and X. index Kr= R/X is my problem that when X is zero it is not finite and i get the error. but i dont know the membership functions of outputs i defined the range of X starts 0.008. Why do i get zero for X?
It seems in first step you have some where integrator and the intial conditions are set to zero. Just check , if R/(x+1) give you the bounded solution. If so then check the initial conditions.
- Badges
- Science topic