Error using wavedec Expected X to be finite. Error in wavedec (line 34) validateat​tributes(x​,{'numeric​'},{'vecto​r','finite​','real'},​'wavedec',​'X');

I am using wavedec to discritize my signal using 'db4' wavelet.

my X = acceleration vector(9000 x 1)

My X is finite. But I am getting a error that ' Error using wavedec Expected X to be finite'

My code :

[a,l] = wavedec(acc,3,'db4');

approx = appcoef(a,l,'db4');

[acd1,acd2,acd3] = detcoef(a,l,[1 2 3]);

figure

subplot(4,1,1)

plot(approx)

title('Approximation Coefficients')

subplot(4,1,2)

plot(acd3)

title('Level 3 Detail Coefficients')

subplot(4,1,3)

plot(acd2)

title('Level 2 Detail Coefficients')

subplot(4,1,4)

plot(acd1)

title('Level 1 Detail Coefficients')

please help me. Where I am wrong. I am unable to solve this problem.

[Carlos Eduardo]

Hola, hermano, perdona por contestar tantos años tarde y en español, el problema puede radicar de dos situaciones, la primera es que el vector es incorrecto (obvio ahí lo dice) pero puede ser ue tengas una columna que enumere tus datos, lo cual es un porblema, el formato .txt ya organiza los datos y la última y es así cómo encontré solución, le tienes que agregar ceros a la izquierda de tus datos, por ejemplo si tienes un 41 y un 1023, lo que tienes que hacer para que los lea como un mismo vector es agregar los respectivos ceros a la izquierda, por ejemplo 00041 y 01023.

Espero le sea de ayuda a alguien más.

Answer 1

[Prasanna]

Hi Arvind,

The error message you are encountering, "Expected X to be finite," indicates that the ‘wavedec’ function is detecting non-finite values (such as NaN or Inf) in your input vector acc. Although you mentioned that your data is finite, you can verify it programmatically using the below code to ensure there are no unexpected values. For Assumption, I am taking a random 9000x1 vector as the acceleration vector.

acc = rand(9000,1); 

% Check for non-finite values 
if any(~isfinite(acc)) 
    error('Input vector contains non-finite values.'); 
end 

% Clean the input vector (if necessary) 
acc = fillmissing(acc, 'linear'); % Replace NaNs with linear interpolation 

% Verify the input vector 
assert(all(isfinite(acc)), 'Input vector must be finite.'); 

[a,l] = wavedec(acc,3,'db4'); 
approx = appcoef(a,l,'db4'); 
[acd1,acd2,acd3] = detcoef(a,l,[1 2 3]); 
figure 
subplot(4,1,1) 
plot(approx) 
title('Approximation Coefficients') 
subplot(4,1,2) 
plot(acd3) 
title('Level 3 Detail Coefficients') 
subplot(4,1,3) 
plot(acd2) 
title('Level 2 Detail Coefficients') 
subplot(4,1,4) 
plot(acd1) 
title('Level 1 Detail Coefficients') 

The above code gives the following as output:

For more documentation on the used functions, you can refer to the following links:

• ‘isfinite’: https://www.mathworks.com/help/releases/R2019b/matlab/ref/double.isfinite.html
• ‘fillmissing’: https://www.mathworks.com/help/releases/R2019b/matlab/ref/fillmissing.html

Hope this helps!