Algorithm 6. Stacking ensemble model.

Result: Prediction of delamination size.

Input: Sensor features with ground truth ${\left(m_p,n_p\right)}_{p=1}^x$.

Output: Stacking ensemble E.

1 Step 1: Develop base-level models CLF on E

2 Perform n-fold cross-validation on base level models

3      clf1 = svr (mp, np)

4      clf2 = ExtraTreeRegressor (mp, np)

5      clf3 = GradientBoostingRegressor (mp, np)

6      clf4 = AdaBoostRegressor (mp, np)

7      clf5 = DecisionTreeRegressor (mp, np)

8 Step 2: Construct the level-on data M

9      $M={\left\{{\hat{m}}_p,n_p\right\}}_{p=1}^x,where$

10      ${\hat{m}}_p$ = {clf1(mp), clf2(mp), clf3(mp), clf4(mp), clf5(mp)}

11 Return E comprising of CLF models