TY - GEN
T1 - Research on Dam Crack Identification Method Based on Multi-source Information Fusion
AU - Xin, Cun
AU - Yang, Dangfeng
AU - Liu, Xiaodong
AU - Huang, Yong
AU - Qian, Xueming
N1 - Publisher Copyright:
© The Author(s) 2025.
PY - 2025
Y1 - 2025
N2 - Cracks as the main safety concern of dams, high-precision identification of dam cracks is of great application value and scientific significance to ensure the safety of dams. The paper proposes a dam crack identification method based on multi-source information fusion. Specifically, image gray scale and geometric features are extracted based on the image information. And then a single crack identification model based on Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), XGBoost, and BP Neural Network are established based on the features, respectively. Finally, a multi-classifier fusion algorithm based on D-S evidence theory is established to identify the presence of cracks by fusing single identification models. Experiments are carried out to compare the proposed method with the existing identification methods based on the evaluation metrics such as accuracy, precision, F1-score, and recall. The results show that the accuracy of crack identification of the proposed method in this paper reaches 98.9%, and the crack identification results are better than the existing methods.
AB - Cracks as the main safety concern of dams, high-precision identification of dam cracks is of great application value and scientific significance to ensure the safety of dams. The paper proposes a dam crack identification method based on multi-source information fusion. Specifically, image gray scale and geometric features are extracted based on the image information. And then a single crack identification model based on Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), XGBoost, and BP Neural Network are established based on the features, respectively. Finally, a multi-classifier fusion algorithm based on D-S evidence theory is established to identify the presence of cracks by fusing single identification models. Experiments are carried out to compare the proposed method with the existing identification methods based on the evaluation metrics such as accuracy, precision, F1-score, and recall. The results show that the accuracy of crack identification of the proposed method in this paper reaches 98.9%, and the crack identification results are better than the existing methods.
KW - Concrete dam
KW - Crack detection
KW - D-S fusion
KW - Machine vision
KW - Multi-information fusion
UR - https://www.scopus.com/pages/publications/85207539404
U2 - 10.1007/978-981-97-9184-2_1
DO - 10.1007/978-981-97-9184-2_1
M3 - 会议稿件
AN - SCOPUS:85207539404
SN - 9789819791835
T3 - Lecture Notes in Civil Engineering
SP - 3
EP - 14
BT - Hydropower and Renewable Energies - Synergistic Integration for Future Energy Systems
A2 - Zheng, Sheng’an
A2 - Wu, Wenhao
A2 - Taylor, Richard M.
A2 - Nilsen, Bjorn
A2 - Zhao, Gensheng
PB - Springer Science and Business Media Deutschland GmbH
T2 - International Hydropower Development Conference, IHDC 2024
Y2 - 31 October 2024 through 31 October 2024
ER -