Genetic Algorithm for parameter estimation in EEMD (Ensemble Empirical Mode Decomposition)

Hello, I have problem how to implement code changes. I have good working EEMD program on a some data. But in the program I have prameter Nstd = 0.3; % param to white noise. which is fixed and it must be adjusting by genetic algorithm (in range 0-1). The EEMD code is bellow:

y=xlsread('some.data');       % load a signal.
aim = 5;                      % numbers of IMF
NR = 10;                      % value of ensemble
Nstd = 0.3;                   % param to white noise

IMF1=eemd(y,aim,NR,Nstd);

and the eemd.m

function [modes] = eemd(y, aim, NR, Nstd)

stdy = std(y);
if stdy < 0.01
    stdy = 1;
end
y = y ./ stdy;

siz = length(y);
modes = zeros(aim+1, siz);
for k = 1:NR
    disp(['Ensemble number #' num2str(k)]);
    wn = randn(1, siz) .* Nstd;
    y1 = y + wn;
    y2 = y - wn;
    modes = modes + emd(y1, aim);
    if Nstd > 0 && NR > 1
        modes = modes + emd(y2, aim);  
    end
end
modes = modes .* stdy ./ (NR);
if Nstd > 0 && NR > 1
    modes = modes ./ 2;
end

end

from emd.m in eemd.m I get the IMF's - as much as is "aim"

[Manal CHOUBI]

Please i would like to have an explication about the two parameters NR and Nstd i dont quiet understand the meaning and the purpose of these parameters , thanks.

[Arpana Singh]

could u plz help me with EEMD . . i m not able to run this code evn . ..if possible mail me at [e-mail]

[Atik Faysal]

@ Arpana Singh, what version of MATLAB are you using? The emd fuction is available from version 2017b and above.

Answer 1

[Kalpesh Prajapati]

Hello, You have to define some objective function such as quality of algorithm (PSNR,MSE etc) as output argument and make standard deviation of noise as input argument as input parameter. Then just apply genetic algorithm on it. For more detail of how to use genetic algorithm, see some example. You can understand it easily.

[Nishtha Rath]

I am doing a project on ensemble empirical mode decomposition, for which I found this MATLAB CODE. I have to use a .wav file as the input. So I converted the .wav into a .mat file. But Here, xlsread() function is used in which I converted my .mat file into its correspondent .xlsx file and tried to run my code with the main function(eemd_test) as:

y=xlsread('Anger.xlsx'); % load a signal.

aim = 5; % numbers of IMF

NR = 10; % value of ensemble

Nstd = 0.3; % param to white noise

IMF1=eemd(y,aim,NR,Nstd);

where, Anger.xlsx is the file name of the .xlsx file I have used in this code. But I couldn't get any output and I get these errors:

Error in eemd (line 14)

modes = modes + emd(y1, aim);

Error in eemd_test (line 5)

IMF1=eemd(y,aim,NR,Nstd);

Please have a look at this and tell me the corrected code in this code.

Thank you!

[ADITYA SHARMA]

I am also getting the same problem.

[Ajeet Patil]

Use below given eend code (Above given code is changed a little bit)

function [modes] = eemd(y, aim, NR, Nstd)

stdy = std(y);

if stdy < 0.01

stdy = 1;

end

y = y ./ stdy;

siz = length(y);

modes = zeros(siz,aim);

for k = 1:NR

disp(['Ensemble number #' num2str(k)]);

wn = (randn(1,siz).*Nstd)';

y1 = y + wn;

y2 = y - wn;

modes = modes + emd(y1,'MaxNumIMF',aim);

if Nstd > 0 && NR > 1

modes = modes + emd(y2,'MaxNumIMF',aim);

end

end

modes = modes .* stdy ./ (NR);

if Nstd > 0 && NR > 1

modes = modes ./ 2;

end

end

[Manal CHOUBI]

Please i would like to have an explication about the two parameters NR and Nstd i dont quiet understand the meaning and the purpose of these parameters , thanks.

[Atik Faysal]

@ Manal Choubi, the NR represents the number of ensembles of EEMD. For each ensemble different white noise is added with the signal. The Nstd represents the white noise that is added. Here 0.3 means the randomly generated white noise has a std of 0.3.

Do some study on the theory of EEMD.

If you are still confused reach me out at [e-mail]

[Jan Ali]

@Ajeet Patil, Could you/anyone please explain what the difference between zeros(siz,aim) and zeros(aim+1, siz) is, why did you put 'MaxNumIMF' in emd(y2,'MaxNumIMF',aim)?

Thanks in advance,