A ship type optimization method based on transfer learning
By employing Two-Stage TrAdaBoost.R2 transfer learning and dual-proxy model optimization algorithms, the problem of high-cost ship model testing in shipbuilding engineering was solved, achieving efficient and accurate ship hull optimization, reducing resource consumption, and improving ship performance.
Patent Information
- Application Number
- CN202310727001.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-10
- Publication Date
- 2026-08-25
- Estimated Expiration
- 2043-08-10
AI Technical Summary
In shipbuilding engineering, traditional ship model testing methods are costly, and obtaining experimental or simulation data is time-consuming and resource-intensive, making it difficult to effectively optimize ship designs to improve performance and efficiency.
We adopted the Two-Stage TrAdaBoost.R2 transfer learning method, combined with a dual-agent model and an improved QPSO optimization algorithm, and used existing experimental data to assist in modeling. We then optimized the ship type through transfer learning when there was insufficient data in the target domain.
It reduces modeling costs, improves model accuracy and optimization speed, ensures the accuracy and reliability of optimization results, is applicable to the optimization of different ship performances, accelerates convergence speed, and reduces the probability of getting trapped in local optima.