Application of ensemble learning algorithm in damage localization of carbon fiber reinforced plastic
In the study of damage localization of Carbon fiber reinforced plastic(CFRP),the localization error of the traditional single model is large.In order to improve the damage positioning accuracy of CFRP,a damage localization method based on ensemble learning algorithm was proposed.The change of material stress caused by heavy load is often used to simulate damage,and different weights are applied to simulate different damage degrees.The experiment is to arrange strain gauges on CFRP to obtain the strain characteristic quantity under different damage degrees.For the same damage degree,the grid search method was used to optimize the parameters of Support vector regression(SVR)model and Classification and regression tree(CART)model.After applying the optimal model parameters to the ensemble learning model,the positioning accuracy of SVR,CART,SVR_Adaboost,random forest(RF),extremely randomized trees(ERT)and Gradient boosting decison tree(GBDT)are compared.The results showed that on the training set of damage degree 1,the mean positioning error of the random forest model and the extreme random tree model with 50 iterations of the weak learner is the lowest,which is 6.2 mm,and was tested on the data sets of damage degree 2 and damage degree 3.The results are indicating that the ensemble learning algorithm can be used on damage localization of CFRP,and the positioning accuracy is higher than that of traditional single models.
NETL
NSTL
万方数据
