Supervisor:
Southwest Ordnance Industry Bureau
Organizer:
Chongqing Ordnance Industry Society
Chongqing University of Technology
Chongqing University of Technology
Research on a quantitative monitoring model for crack propagation based on generalized regression neural network
| DOI: | 10.11809/bqzbgcxb2025.01.026 |
| Keywords: | crack propagation; neural network; sensor; feature extraction; data fusion |
| Abstract: | Cracks formed in the key components of armored vehicles can lead to performance degradation and even accidents, thus requiring the development of advanced crack monitoring techniques. This paper proposes a crack propagation quantitative monitoring model based on generalized regression neural network, which can predict the crack length and propagation rate according to the resistance strain signal and Lamb wave signal generated by the structure under cyclic load. A multi sensor crack propagation feature extraction method is adopted, which uses two different types of sensors, piezoelectric sensor and resistance strain gauge, and combines passive and active monitoring modes to extract the feature parameters related to crack length and propagation rate from the resistance strain signal. Two different data processing techniques, random forest algorithm and D S evidence theory, are used to achieve effective identification and data fusion of crack length. Based on the resistance strain signal and Lamb wave signal under different crack lengths, different load frequencies, and different sensor positions, the model training and testing are carried out. |
| Issue: | Vol. 46 No. 1 (2025) |
| Published: | 2025-01-31 |