Details

Autor(en) / Beteiligte

• Sun, Yiping
• Tao, Meng

Titel

Convolutional Neural Network Prediction of Underwater Anechoic Coating: Effect of Material Properties on Absorption Coefficient

Ist Teil von

• Proceedings of the World Conference on Intelligent and 3-D Technologies (WCI3DT 2022), p.251-259

Ort / Verlag

Singapore: Springer Nature Singapore

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• A cavity viscoelastic structure is used as an underwater anechoic coating because of its excellent acoustic performance. To overcome the shortcomings of the traditional numerical analysis method, which is time-consuming, and physical experiment for measuring the absorption coefficient is costly, a convolutional neural network-based (CNN-based) logistical prediction method was investigated. First, the finite element method (FEM) was used to calculate the absorption coefficient. Then, a dataset was constructed based on the FEM; in this way, it avoided repeated experiments. The critical material parameters were processed by the min–max normalization method, and the ground truth of the dataset is the absorption coefficient. Finally, the root-mean-square error (RMSE) predicted by the CNN-based model was 0.1177, and the Pearson correlation coefficient (PCC) was 0.9354. It shows that the CNN-based logistic regression model accurately and rapidly predicts the effect of material properties on the absorption coefficient of the underwater viscoelastic anechoic coating.