Algorithm 3. 2LE-CEEMDAN denoising method.

Input: Timeseriesdata

$IMFs:[IM{F}_0,IM{F}_1,...,IM{F}_{k-1}]\leftarrow$ k IMFs obtained with CEEMDAN() (Algorithm 1)

procedure SELECTION_IMF(IMFs)

$sample\mathrm{\_}entropy\mathrm{\_}list=[]$

$approximate\mathrm{\_}entropy\mathrm{\_}list=[]$

for imf in IMFs do

$sample\mathrm{\_}entropy\mathrm{\_}list.append(SampEn(imf))$

$approximate\mathrm{\_}entropy\mathrm{\_}list.append(ApEn(imf))$

$total\mathrm{\_}SampEn=sum(sample\mathrm{\_}entropy\mathrm{\_}list)$

$total\mathrm{\_}ApEn=sum(approximate\mathrm{\_}entropy\mathrm{\_}list)$

$sample\mathrm{\_}entropy\mathrm{\_}ratio$ = $\left[\frac{entropy*100}{total\_SampEn}\text{for\hspace{0.17em}entropy\hspace{0.17em}in\hspace{0.17em}}{\text{sample }}_{\text{e}}\text{ntropy\hspace{0.17em}list}\right]$

$approximate\mathrm{\_}entropy\mathrm{\_}ratio$ = $\left[\frac{entropy*100}{total\_ApEn}\text{\hspace{0.17em}for entropy in approximate entropy list}\right]$

end for

$noisy\mathrm{\_}imfs=[]$

$noiseless\mathrm{\_}imfs=[]$

for i in range(len(IMFs)) do

if $sample\mathrm{\_}entropy\mathrm{\_}ratio[i]>20$ $approximate\mathrm{\_}entropy\mathrm{\_}ratio[i]>20$ then

$noisy\mathrm{\_}imfs.append(IMFs[i])$

else

$noiseless\mathrm{\_}imfs.append(IMFs[i])$

end if

end for

return $noisy\mathrm{\_}imfs$, $noiseless\mathrm{\_}imfs$

end procedure

$First\mathrm{\_}IMFs=CEEMDAN(Input)$

$First\mathrm{\_}decomposition\mathrm{\_}noisy\mathrm{\_}IMF,First\mathrm{\_}decomposition\mathrm{\_}noiseless\mathrm{\_}IMF=Selection\mathrm{\_}IMF(First\mathrm{\_}IMFs)$

$noisly\mathrm{\_}First\mathrm{\_}IMFs\mathrm{\_}sum=sum(First\mathrm{\_}decomposition\mathrm{\_}noisy\mathrm{\_}IMF)$

$Second\mathrm{\_}IMFs=CEEMDAN(noisly\mathrm{\_}First\mathrm{\_}IMFs\mathrm{\_}sum)$

$Second\mathrm{\_}decomposition\mathrm{\_}noisy\mathrm{\_}IMF,Second\mathrm{\_}decomposition\mathrm{\_}noiseless\mathrm{\_}IMF=Selection\mathrm{\_}IMF(Second\mathrm{\_}IMFs)$