First try to understand what singularity is? It is the break down of the computation process at somewhere due to the numerical conditions. For instance if there is a division by 0 or may be trigonometric infinity solutions. Here 'NaN' is the breakpoint.

To solve these type of error, you can go through following steps:

1. Use the Simulink debugger (in a model window, Simulation > debug > debug model) to identify where is the breakpoint (Inf or NaN) originates. ( Details are here: http://www.mathworks.com/help/simulink/ug/starting-the-simulink-debugger.html )

OR

2. Use the unit delay block / memory block.

OR

3. If you are integrating with zero initial condition in integrator block try 0.0000001 instead. (it works for sometimes.)

OR

4. Try reducing fixed step size.

Try these steps :-

1. In simulation-->configuration parameters-->solver-->select "Non-adaptive". Then try simulating. If not successful go to step-2.

2. simulation-->configuration parameters-->solver-->increase the value "relative tolerance" & "absolute tolerance" in the range of 1e-4 to 1e-6. Then try simulating.

Please let me know if it worked.

Thanks in advance.

Change the step to avoid going over poles of the system.

Hi,

Often, it is indicating that your model is instable which can bedue to something as trivial as having a wrong sign in a gain or a sum block.

In addition to the previous pieces of advice, note that the solver can bealso an issue. If your system has very rapidly varying dynamics or discontinuities for instance. So you might want to try other solvers such as ode15s, ode23s or ode23tb.

Hi,

In addition to what has been previously said, try reducing the amplitudes of the inputs

Am experiencing the same problem 

tried by increasing the values "relative tolerance" & "absolute tolerance in solver tabs of the config parameters but the problem persists.

Please put in a zero order delay to resolve this. An "algebraic" loop is the natural consequence of flow graph execution in time. An additional constraint is needed to remove the ambiguity "what should be computed first at a time step" much like the "chicken-egg" paradox. The unit delay will force one part of the loop to look back in time and use the previous value. Graphical representation and execution are not the same thing.

this has worked for me

First try to understand what singularity is? It is the break down of the computation process at somewhere due to the numerical conditions. For instance if there is a division by 0 or may be trigonometric infinity solutions. Here 'NaN' is the breakpoint.

To solve these type of error, you can go through following steps:

1. Use the Simulink debugger (in a model window, Simulation > debug > debug model) to identify where is the breakpoint (Inf or NaN) originates. ( Details are here: http://www.mathworks.com/help/simulink/ug/starting-the-simulink-debugger.html )

OR

2. Use the unit delay block / memory block.

OR

3. If you are integrating with zero initial condition in integrator block try 0.0000001 instead. (it works for sometimes.)

OR

4. Try reducing fixed step size.

excellent Patel.. You have summed it up all

you can try simulink debugger. it help me to trace the NaN dan zero crossing

you can use Saturation block to fix the prob

I used a saturation block but it doesn't run

If you have an "algebraic loops" in your Simulink model, you can use Memory bolok.

For more details, you can see this link "http://www.mathworks.com/help/simulink/ug/algebraic-loops.html"

You have to look into the inputs of the integrator block to understand if it is going for NaN or for Inf.

You get NaN if for example you divide 0/0, you get Inf if you divide a number other than zero by zero.

If the input is Inf, that is solvable by a saturation block, as it was described above. Since you say it does not solve the problem, I would be guessing for a NaN as an input, meaning that if you're trying to integrate let's say f(x)=1/x , you have to make sure to make an exception for when x==0, for example with an if block.

Please refer to: https://mathworks.com/help/matlab/ref/nan.html

it is very much useful and my error got rectified

@ Ricardo Baila thanks for your answer. it is really helpful

I am facing similar issues . please suggest Saturation or unit delay block where to connect in simulink model

Sometimes the problem is that the integrator is not converging. Go to Model Configuration Parameters -> Solver and choose a variable step solver. Usually ode45 works (do not try auto). Then, choose suitable values for Max step size, Min. step size, Initial step size, relative tolerance and absolute tolerance.

When an error such as yours happen, reducing max, min and initial step size usually work.

Samia Larbi Is that you have found a solution for this problem ?

Derivative of state '1' in block 'XX' at time 0.00445 is not finite. The simulation will be stopped. 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) ...

how can i solve it please ?????

First try to solve it as explained by @Bhavik Patel and @Ricardo Baila. Moreover, you can try to change the solver as Vinícius Gabriel Segala Simionatto suggested. If does not work, you can try the solution described as below.

Create a Matlab function:

function y = fcn(u)

% Verify that the input is not a number. If it is true, the function output will be made equal to 0 or eps. This can be necessary when a signal passes through a function that uses an integral/derivative.

if isnan(u) || isinf(u)

u=0; % Or eps (to avoid division by zero)

end

y = u;

Remember to make the function as "Treat as atomic unit".

Another solution is try to use the block "IC" (Initial condition) to solve this problem.

You can provide initial value to the variable which appears in the denominator in SIMULINK. See the attachment. I hope , it will work for you.

try using a first order filter having transfer function as ''1/(0.1*s+1)" at the output of derivative block. This will solve your problem.

Hi. You can add saturation block and set an appropriate range for it.