Multi-condition optimisation design of a hydrofoil based on deep belief network. (15th March 2023)

• Record Type: Journal Article
• Authors: Zhu, Guojun; Feng, Jianjun; Li, Ping; Wang, Zhaoning; Wu, Guangkuan; Luo, Xingqi
• Is Part Of: Ocean engineering. Volume 272(2023)
• Publication Date: 2023-03-15
• Subjects: Multi-condition -- Optimisation design -- Hydrofoil -- Deep belief network; Ocean engineering -- Periodicals; Ocean engineering; Periodicals; 620.4162
• DOI: 10.1016/j.oceaneng.2023.113846
• Languages: English
• ISSNs: 0029-8018

Abstract

Abstract: A hydrofoil optimisation method based on the combination of deep belief network (DBN) and non-dominated sorting genetic algorithm II (NSGA-II) is proposed in this paper. Firstly, the number of objective functions in the multi-condition optimisation of hydrofoil is reduced by the performance distance formula. Then, Multi-DBN is adopted to replace computational fluid dynamics (CFD) simulation to establish the response relationship between geometric parameters and hydrodynamic performance of the hydrofoil. Moreover, the network parameters of Multi-DBN are optimised to improve prediction accuracy. Finally, the hydrofoil optimisation method is established by combining Multi-DBN and NSGA-II algorithms. Based on this method, the hydrodynamic performances of the hydrofoil NACA 63–815 at angles of attack (AoA) of 0°, 6° and 12° are optimised. Results indicate that the performances of the optimised hydrofoil are better than the initial at the three AoAs. Moreover, within the AoA range between 0° and 14°, the maximum improvements of the lift–drag ratio and the minimum pressure coefficients are 10.63% and 11.44%, respectively. Using Multi-DBN instead of CFD as the performance evaluation method in optimisation can greatly reduce the time of optimisation design. Highlights: A Multi-DBN model is proposed to establish the response relationship between hydrofoil geometry and performance. A method based on performance distance is proposed to reduce the number of objective functions. The presented optimisation method has shortened the time of hydrofoil optimisation.