lsim giving NaN output for zero input

Hi,

I have a system with frequency and phase response as shown. The system is unstable with one pole on the right half of s-plane. If input is x=[0 0 0] output through lsim is NaN. How can this issue be solved ? Please help. Thank you.

Zeros=(-37436128.4318415 + 29344189.3569776i , -37436128.4318415 - 29344189.3569776i , -4398673.13805009 + 4099882.75353130i , -4398673.13805009 - 4099882.75353130i, -632969.549556042 + 0.00000000000000i)

Poles=( -1105299964.29975 + 0.00000000000000i, -102844523.541346 + 0.00000000000000i, -23575505.1105442 + 30425526.2059357i, -23575505.1105442 - 30425526.2059357i, 24762584.8470421 + 0.00000000000000i)

Gain k = -8.8777e+04

Answer 1

[Raj]

I am getting zero output for zero input if that's what you are looking for. Check how you are using lsim.

Zeros=[-37436128.4318415 + 29344189.3569776i  , -37436128.4318415 - 29344189.3569776i , -4398673.13805009 + 4099882.75353130i ,  -4398673.13805009 - 4099882.75353130i,  -632969.549556042 + 0.00000000000000i];
Poles=[ -1105299964.29975 + 0.00000000000000i, -102844523.541346 + 0.00000000000000i, -23575505.1105442 + 30425526.2059357i,  -23575505.1105442 - 30425526.2059357i,  24762584.8470421 + 0.00000000000000i];
k = -8.8777e+04;
sys=zpk(Zeros,Poles,k) % Define your system
t=0:0.1:1; % lets say we run the simulation for 1 second
x=zeros(1,numel(t)); % all zero input till end of simulation
lsim(sys,x,t)

[shubha t r]

Zeros=[-37436128.4318415 + 29344189.3569776i , -37436128.4318415 - 29344189.3569776i , -4398673.13805009 + 4099882.75353130i , -4398673.13805009 - 4099882.75353130i, -632969.549556042 + 0.00000000000000i];

Poles=[ -1105299964.29975 + 0.00000000000000i, -102844523.541346 + 0.00000000000000i, -23575505.1105442 + 30425526.2059357i, -23575505.1105442 - 30425526.2059357i, 24762584.8470421 + 0.00000000000000i];

k = -8.8777e+04;

sys=zpk(Zeros,Poles,k) % Define your system

t=0:0.1:1; % lets say we run the simulation for 1 second

x=zeros(1,numel(t)); % all zero input till end of simulation

[y,t1]=lsim(sys,x,t)

plot(t1,y);

Here y is NaN and there is no plot.

[Raj]

NaN doesn't mean error. There may be many reasons for getting a NaN output. As a way around here, I suggest using 'filter' command instead of 'lsim'. Something like this:

Zeros=[-37436128.4318415 + 29344189.3569776i  , -37436128.4318415 - 29344189.3569776i , -4398673.13805009 + 4099882.75353130i ,  -4398673.13805009 - 4099882.75353130i,  -632969.549556042 + 0.00000000000000i];
Poles=[ -1105299964.29975 + 0.00000000000000i, -102844523.541346 + 0.00000000000000i, -23575505.1105442 + 30425526.2059357i,  -23575505.1105442 - 30425526.2059357i,  24762584.8470421 + 0.00000000000000i];
k = -8.8777e+04;
sys=zpk(Zeros,Poles,k) % Define your system
%t=0:0.1:1; % lets say we run the simulation for 1 second
dt = 1/10; 
t = dt*(0:10); 
x=zeros(1,numel(t)); % all zero input till end of simulation
%lsim(sys,x,t)
[num den]=tfdata(sys)
y=filter(cell2mat(num),cell2mat(den),x)
plot(t,y);

Hope this helps!!

[shubha t r]

Thank you.

Can you please explain the difference betweeen lsim and filter? Is filter only for discrete time systems?

[Raj]

I have edited my previous answer a bit. Please take care of that.

You are right about point that filter is for discrete time systems. For details see

https://in.mathworks.com/help/control/ref/lsim.html#f3-209209

and

https://in.mathworks.com/help/matlab/ref/filter.html

[shubha t r]

Thanks a lot:)