Algorithm 1. CEEMDAN decomposition method.

Output: $\widetilde{IM{F}_k}$

$r_k(n)\leftarrow residues$

$E_j(.)\leftarrow j$-th IMF obtained by EMD decomposition

$w^i\leftarrow$ White noise

$x(n)\leftarrow$ Time series signal

${\epsilon }_0\leftarrow$ Noise coefficient

$I\leftarrow$ Number of trials

$IM{F}_1^i(n)=x(n)+{\epsilon }_0{w}^i(n)$

$\widetilde{IM{F}_1}(n)=0$

for $i=1$ to I do

$\widetilde{IM{F}_1}(n)=\widetilde{IM{F}_1}(n)+IM{F}_1^i(n)/I\leftarrow$ First IM

end for

$r_1(n)=x(n)-\widetilde{IM{F}_1}(n)\leftarrow$ First residue

$\widetilde{IM{F}_2}(n)=0$

for $i=1$ to I do

$\widetilde{IM{F}_2}(n)=\widetilde{IM{F}_2}(n)+E_1(r_1(n)+{\epsilon }_1{E}_1(w^i(n)))/I\leftarrow$ Second IMF

end for

while $r_k(n)\leftarrow$ until the value of residual component is less than two extremes do

for $k=2$ to K do

$r_k(n)=r_{k-1}(n)-\widetilde{IM{F}_k}(n)\leftarrow$ Residuals for $k=1,2,..K$

$\widetilde{IM{F}_k}(n)=0$

for $i=1$ to I do

$\widetilde{IM{F}_{k+1}}(n)=\widetilde{IM{F}_{k+1}}(n)+E_1(r_k(n)+{\epsilon }_k{E}_k(w^i(n)))/I$

end for

end for

end while