Unlock instant, AI-driven research and patent intelligence for your innovation.

A Surrogate Model Experiment Design Method Based on Hypervolume Iterative Strategy

A surrogate model and experimental design technology, applied in computing, electrical digital data processing, special data processing applications, etc., can solve problems such as difficulty in predicting optimal solutions for surrogate models, affecting the application effect of surrogate models, and lack of sample selection criteria. The effect of improving limitations, reducing economic costs, and reducing the amount of calculation

Active Publication Date: 2019-01-08
BEIHANG UNIV
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At present, the commonly used method is to take a certain multiple of the dimension of the design space as the sample size, which lacks consideration of other factors, and lacks relevant sampling criteria to guide the establishment of effective samples, which makes it difficult for the surrogate model to predict the optimal solution and directly affects The application effect of the proxy model

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Surrogate Model Experiment Design Method Based on Hypervolume Iterative Strategy
  • A Surrogate Model Experiment Design Method Based on Hypervolume Iterative Strategy
  • A Surrogate Model Experiment Design Method Based on Hypervolume Iterative Strategy

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0048] In order to understand more fully the characteristics of the invention and its applicability to engineering practice, the present invention aims at such as image 3 The one-dimensional function shown constructs the surrogate model. This kind of one-dimensional function has the following characteristics: in a given interval, there are not only areas with relatively sharp changes, but also areas with relatively gentle changes. It is obvious that the traditional uniform distribution method can not completely fit the corresponding model. In the case of fewer sample points, the required optimal solution may even be lost. By changing the allowable tolerance, we selected 7 and 9 sample points respectively according to the integration point arrangement scheme introduced in the invention to construct the proxy model. available from Figure 4 and Figure 5 It can be seen intuitively that compared with the traditional evenly distributed point scheme, the sample points selected ...

Embodiment 2

[0050] The finite element model of a certain wing box section is shown in the figure, the shell elements pshell.23 and pshell.24 are selected, and their thickness t 8 , t 9 is the variable, and the wingtip deflection w is the response, and the HVI method is used to select sample points to build a proxy model. Select the points in the table as the verification data, and compare the wingtip deflection directly obtained through the proxy model with the deflection obtained by the actual finite element calculation. Through the error column in the table, it can be clearly found that the constructed proxy model has fewer sample points In the case of , it still has enough precision.

[0051] To sum up, the present invention proposes a proxy model construction method based on hypervolume iterative strategy. Firstly, the initial area and allowable tolerance are given according to the actual engineering needs and design experience; secondly, using the hypervolume iterative strategy, co...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a delegation-model experiment design method based on the hyper-volume iteration (HVI) strategy. The design result obtained with the method better meets the actual requirements, and the engineering applicability is higher. A new strategy based on 'a hyper-volume difference transverse-moving datum line thought' is firstly provided, and experiment design, namely, the selection method of a sample point in the delegation-model establishment process is improved. Compared with the traditional uniform-distribution point distributing or gauss integration point thought, a new self-adaption integration point arrangement method based on the simpson formula is characterized in that for increasing the accuracy, the calculated amount is considered at the same time, many integration points are selected at a place with the drastic function change, and small integration points are selected at a place with the flat function change. The delegation-model experiment design method has the advantages that according to the specific characteristics of a target function, intelligent integration point distribution can be carried out, and therefore the computational accuracy problem caused by the subjectivity is effectively solved, or the redundancy problem of the integration points is solved.

Description

technical field [0001] The invention relates to the technical field of experiment design in the process of agent model construction, in particular to an agent model experiment design method based on a hypervolume iterative strategy. Background technique [0002] The optimization of large-scale design parameters is often encountered in the design of aircraft. For example, the aerodynamic design of the wing of a large aircraft needs to optimize about 100 design parameters, and the design parameters for the whole aircraft can reach thousands to tens of thousands. The optimization of large-scale design parameters often requires larger-scale numerical calculations, and the optimization efficiency of direct calculations is very low in terms of current computing power. In order to solve the optimization design of large-scale parameters in aircraft design, on the one hand, it is necessary to continue to strengthen investment in the construction of high-performance computing capabili...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F17/50
CPCG06F30/15
Inventor 王晓军倪博文王磊刘易斯罗振先刘冠华
Owner BEIHANG UNIV