Hull local curved surface optimization neural network modeling method and hull local curved surface optimization method

Abstract

The invention discloses a ship body local curved surface optimization neural network modeling method and a ship body local curved surface optimization method, and relates to the field of ship design. In order to solve the problem that a large amount of initial sample data often need to be input in an existing agent model, the technical scheme provided by the invention is as follows: the method comprises the following steps: selecting a plurality of control points in a to-be-optimized ship body; changing the coordinates of the control points to obtain new control points, and taking the new control points as sample data corresponding to a plurality of ship types; hydrodynamic characteristic values are obtained to construct a BP neural network model; collecting a ship hydrodynamic characteristic value calculated by a BP neural network model and a difference value of ship hydrodynamic characteristic values corresponding to the group of control point coordinates obtained by a CFD technology as an error value; establishing a Kriging model, and taking an error value as output data; taking the sum of the error value and the output data as judgment data; and performing judgment, and if the judgment data meets a preset requirement, taking the judgment data as output data in combination with a BP neural network model and a Kriging model, thereby completing training. The method is suitable for being applied to ship body curved surface design work.

Description

technical field [0001] It involves the field of ship design, specifically the field of ship hull shape design. Background technique [0002] Neural network is a commonly used method in ship hydrodynamic performance prediction. Among them, BP neural network is used more, but BP neural network needs more samples to achieve a certain accuracy, and it is impossible to accurately select samples when the samples need to be updated. [0003] The optimization of the hull surface is one of the important means to realize the energy saving of the ship. The optimization of the hull surface is mainly to optimize the hydrodynamic performance of the ship. In the process of optimizing the hydrodynamic performance of the ship, the prediction of the ship's resistance performance is very important. And ship resistance is an important factor affecting the design of ships, which is directly related to the technical and economic indicators of ships. At present, the prediction of ship resistance ...

AI Technical Summary

Problems solved by technology

[0006] For the existing BP neural network, the samples can only be updated by randomly increasing the initial sample data, and it is impossible to accurately select samples that can improve the prediction accuracy, resulting in slow convergence and weakening the superiority of using the BP neural network model for ship type optimization. , the technical solution provided by the invention is:

Images