Application of Machine Learning Algorithms for Tool Condition Monitoring in Milling Chipboard Process

View full-text article in PMC

. 2023 Jun 23;23(13):5850. doi: 10.3390/s23135850

Algorithm 2 CalcFeatures_cwt function

1:
procedure CalcFeatures_cwt( $x, w a v e l e t n a m e, l e v e l$ )
2:
$feature_levels \leftarrow empty list$
3:
for $i i \in {0, 1, \dots, l e v e l - 1}$ do
4:
$(x, coeff_d) \leftarrow DWT (x, w a v e l e t n a m e)$
5:
Calculate statistical features for x:
- $w a v e_y 1 \leftarrow mean (x)$
- $w a v e_y 2 \leftarrow \max (x)$
- $w a v e_y 3 \leftarrow RMS (x)$
- $w a v e_y 4 \leftarrow std (x)$
- $w a v e_y 5 \leftarrow std (x) / mean (x)$
- $w a v e_y 6 \leftarrow \max (x) / mean (x)$
6:
Calculate statistical features for $coeff_d$ :
- $w a v e_c o e f_y 1 \leftarrow mean (coeff_d)$
- $w a v e_c o e f_y 2 \leftarrow \max (coeff_d)$
- $w a v e_c o e f_y 3 \leftarrow RMS (coeff_d)$
- $w a v e_c o e f_y 4 \leftarrow std (coeff_d)$
- $w a v e_c o e f_y 5 \leftarrow std (coeff_d) / mean (coeff_d)$
- $w a v e_c o e f_y 6 \leftarrow \max (coeff_d) / mean (coeff_d)$
- $w a v e_c o e f_n 5 \leftarrow 5 % - p e r c e n t i l e (coeff_d)$
- $w a v e_c o e f_n 25 \leftarrow 25 % - p e r c e n t i l e (coeff_d)$
- $w a v e_c o e f_n 75 \leftarrow 75 % - p e r c e n t i l e (coeff_d)$
- $w a v e_c o e f_n 95 \leftarrow 95 % - p e r c e n t i l e (coeff_d)$
  7:
         $features \leftarrow (w a v e_y 1, w a v e_y 2, w a v e_y 3, w a v e_y 4, w a v e_y 5, w a v e_y 6,$

                $w a v e_c o e f_y 1, w a v e_c o e f_y 2, w a v e_c o e f_y 3, w a v e_c o e f_y 4,$

                $w a v e_c o e f_y 5, w a v e_c o e f_y 6, w a v e_c o e f_n 5, w a v e_c o e f_n 25, w a v e_c o e f_n 75,$

                $w a v e_c o e f_n 95)$
8:
Append $features$ to $feature_levels$
9:
end for
10:
$output \leftarrow Flatten (feature_levels)$
11:
return $output$
12:
end procedure