|
|
|
| A STUDY ON CONSTRAINT-RELATED FRACTURE TOUGHNESS PREDICTION BASED ON RANDOM FOREST ALGORITHM AND DATA ENHANCEMENT STRATEGY |
|
|
|
|
Abstract In this study, the nuclear power steel A508 was selected, and the prediction ability of K-nearest neighbor regression (KNN), nuclear regression (KR), linear regression (LR) and random forest regression (RF) was investigated. It was found that the prediction effect of the four algorithms on the constraint-related fracture toughness is RF > LR > KNN > KR. Further, based on the RF algorithm, the constraint-related fracture toughness of different specimens under different constraints was predicted, and the data under plane stress and plane strain were added to the data for data enhancement. The results show that by adding data enhancement strategy, the fracture toughness prediction model is further improved, the prediction results are more accurate, and the trained model has better generalization ability.
|
|
Received: 10 July 2024
Published: 28 February 2025
|
|
|
|
|
|
|
|
|
2025, Vol. 46 
