Search

CN-121997447-A - Hydrofoil proxy model optimization method and hydrofoil ship

CN121997447ACN 121997447 ACN121997447 ACN 121997447ACN-121997447-A

Abstract

The invention provides a hydrofoil proxy model optimization method and a hydrofoil ship, wherein the hydrofoil proxy model optimization method comprises the steps that a two-stage addition point Kriging proxy model is used for accurately predicting hydrofoil performance, then NSGAII algorithm is used for hydrofoil optimization, the calculation efficiency is ensured, meanwhile, the performance prediction precision of the hydrofoil strong nonlinear working condition is obviously improved, and global optimization and local refinement dynamic balance are realized.

Inventors

  • Hu Wulong
  • CHENG YONGFENG
  • XING MENGYAO

Assignees

  • 武汉理工大学三亚科教创新园
  • 武汉理工大学

Dates

Publication Date
20260508
Application Date
20251215

Claims (10)

  1. 1. The hydrofoil proxy model optimization method is characterized by comprising the following steps of: s1, determining the design variable number of the hydrofoil and the value range of each design variable, wherein all the design variables take values in the respective value ranges, and mutually combining to construct a sample space of the hydrofoil; S2, calculating performance response values corresponding to the hydrofoils respectively based on the training samples and the test samples; s3, constructing a first Kriging agent model by using the training sample and the performance response value thereof; s4, verifying the accuracy of the first Kriging agent model by using the test sample and the performance response value thereof; S5, if the precision reaches the standard, performing multi-objective optimization on the first Kriging agent model by adopting a multi-objective genetic algorithm to obtain a pareto front edge, and selecting a final optimal solution from the pareto front edge by using a double-base-point method; And S6, if the precision does not reach the standard, correcting the first Kriging proxy model by adopting a double-stage dotting strategy, wherein the double-stage dotting strategy comprises a primary dotting based on an expected lifting criterion and a secondary dotting based on a prediction variance criterion, wherein the sample is obtained from the sample space, added into the training sample, returned to the steps S2 to S4 until the precision reaches the standard, and then operated in the step S5.
  2. 2. The hydrofoil proxy model optimization method of claim 1, it is characterized in that the method comprises the steps of, In step S1, the design variables include the angle of attack, depth of immersion, and speed of the hydrofoil.
  3. 3. The hydrofoil proxy model optimization method of claim 2, it is characterized in that the method comprises the steps of, In step S2, the performance response values include lift coefficient, drag coefficient, and lift-drag ratio of the hydrofoil.
  4. 4. The hydrofoil proxy model optimization method of claim 1, it is characterized in that the method comprises the steps of, In step S6, the primary adding point based on the desired lifting criterion includes: s6.1.1, constructing a second kriging agent model based on all samples of the sample space; s6.1.2, acquiring a sample with the maximum expected lifting value from the second kriging agent model by adopting a genetic algorithm, and taking the sample out of the second kriging agent model; s6.1.3 repeating S6.1.2, setting the repetition times as a third preset value, and constructing the second kriging agent model for taking out all the samples as a third kriging agent model, wherein all the taken out samples form a design point library.
  5. 5. The hydrofoil proxy model optimization method of claim 4, it is characterized in that the method comprises the steps of, In step S6.1.2, a genetic algorithm is adopted to obtain a sample with the maximum expected lifting value from the second kriging agent model, the initial population of decision variables of the genetic algorithm is set to 180-220, and the iteration number is set to 100-140.
  6. 6. The method for optimizing a hydrofoil proxy model according to any one of claims 1 to 5, In step S6, the secondary adding point based on the prediction variance criterion includes: S6.2.1, processing samples in the design point library by adopting a genetic algorithm to simulate a binary crossover operator to generate a secondary candidate point library; S6.2.2, calculating the mean square error of all samples of the secondary candidate point library based on the third kriging proxy model, taking the sample with the maximum value in the mean square error as a final sample, and adding the final sample into the training sample.
  7. 7. The hydrofoil proxy model optimization method of claim 6, it is characterized in that the method comprises the steps of, In step S6.2.2, the process of obtaining the sample of the maximum value in the mean square error includes: and taking samples with larger fourth preset values from the mean square errors of all the samples to form a fourth Kerling proxy model, and acquiring the sample with the largest mean square error by adopting a multi-target genetic algorithm as a final sample.
  8. 8. The hydrofoil proxy model optimization method of claim 7, it is characterized in that the method comprises the steps of, In step S1, the preset number of training samples is set to a first preset value, the preset number of test samples is set to a second preset value, and the first preset value is greater than the second preset value.
  9. 9. The hydrofoil proxy model optimization method of claim 8, it is characterized in that the method comprises the steps of, The design variable number is set to be n, n is more than or equal to 3, the first preset value is set to be 45-60 n, the second preset value is set to be 8-12 n, and/or, The third preset value is set to 20-30, and/or, The fourth preset value is set to 13-18.
  10. 10. Hydrofoil vessel comprising a hydrofoil obtained according to the hydrofoil proxy model optimization method of any of claims 1-9.

Description

Hydrofoil proxy model optimization method and hydrofoil ship Technical Field The invention relates to the technical field of hydrofoil vessels, in particular to a hydrofoil proxy model optimization method and a hydrofoil vessel. Background The hydrofoil of the hydrofoil ship is an energy-saving hydrodynamic device for generating lift force based on the Bernoulli principle, the sailing resistance of the ship is obviously reduced by lifting the hull of the hydrofoil ship to be above the water surface, and the hydrofoil is a core component part of the hydrofoil ship. Currently, hydrofoil performance optimization relies mainly on two approaches, computational Fluid Dynamics (CFD) simulation and physical experiments. However, CFD-based optimization design, especially in global optimization involving multiple operating conditions and multiple design variables, is particularly problematic in the preliminary design and parameter sensitivity analysis stage due to the large number of numerical simulations that need to be performed, with a large computational burden. The physical method is subject to high cost and long period, and is difficult to support for rapid iteration. The approximate mapping relation between the design parameters and the performance response is constructed by adopting a proxy model, and the approximate mapping relation is applied to hydrofoil performance optimization by combining global optimization strategies such as an evolutionary algorithm. The framework takes a limited high-fidelity CFD sample as a reference, and realizes the rapid performance estimation of a large-scale design scheme through a proxy model, so that the optimization period is obviously shortened. Although the method relieves the pressure of the calculation cost to a certain extent, the existing proxy model technology has the following key defects when processing the problems of strong nonlinearity and multi-parameter coupling flow in the hydrofoil system: 1) The traditional agent model generally adopts a one-time static sampling strategy, and after the model is constructed, an active learning and updating mechanism for a sample space is lacked, so that the fitting capacity of the model in a strong nonlinear region is weak, and the prediction precision and the modeling efficiency are difficult to be compatible. 2) The existing optimization framework lacks a systematic self-adaptive sampling guiding strategy, can not dynamically coordinate global exploration and local development in the optimization process, and restricts the capability and robustness of an algorithm for finding a global optimal solution. Therefore, a hydrofoil proxy model performance optimization scheme is required to be invented, and the problems of insufficient prediction precision and efficiency existing in the conventional hydrofoil proxy model performance optimization are solved. Disclosure of Invention The invention provides a hydrofoil proxy model optimization method and a hydrofoil ship, which can overcome the difficulty that the proxy model prediction precision and efficiency are insufficient in the hydrofoil optimization method in the prior art. In order to solve the problems, the invention provides a hydrofoil proxy model optimization method, which comprises the following steps: s1, determining the design variable number of the hydrofoil and the value range of each design variable, wherein all the design variables take values in the respective value ranges, and mutually combining to construct a sample space of the hydrofoil; S2, calculating performance response values corresponding to the hydrofoils respectively based on the training samples and the test samples; s3, constructing a first Kriging agent model by using the training sample and the performance response value thereof; s4, verifying the accuracy of the first Kriging agent model by using the test sample and the performance response value thereof; S5, if the precision reaches the standard, performing multi-objective optimization on the first Kriging agent model by adopting a multi-objective genetic algorithm to obtain a pareto front edge, and selecting a final optimal solution from the pareto front edge by using a double-base-point method; And S6, if the precision does not reach the standard, correcting the first Kriging proxy model by adopting a double-stage dotting strategy, wherein the double-stage dotting strategy comprises a primary dotting based on an expected lifting criterion and a secondary dotting based on a prediction variance criterion, wherein the sample is obtained from the sample space, added into the training sample, returned to the steps S2 to S4 until the precision reaches the standard, and then operated in the step S5. According to the technical scheme, a sample space of the hydrofoil is constructed through design variables of the hydrofoil, a part of samples are selected to serve as a model for constructing a Kriging proxy, the model is verified by the other part, the p