An Optimization Process Modeling Method Supporting Response Surface Estimation and Update

Abstract

The invention discloses an optimization process modeling method supporting response surface estimation and updating, comprising the following steps: using an experimental design component to obtain an initial design point, and importing the above initial design point as an input variable into an expression component to obtain an output variable; Use the response surface component to construct the initial response surface based on the above input variables and output variables; realize the estimation calculation of the initial response surface by referencing the component, and calculate the optimal point of the initial response surface; load the optimal point into the expression as a new input variable In the component, a new output variable is generated; a new response surface is generated based on the new input variable and output variable, and the response surface is updated; a new optimal point is calculated based on the new response surface; and the optimal point is converged according to the new optimal point Judgment and obtain the optimal value, so as to realize the modeling of the optimization process. The invention uses the test design component, the response surface component and the reference component in combination to realize the dynamic update and estimation of the response surface, and has high modeling efficiency.

Description

technical field [0001] The invention belongs to the technical field of multidisciplinary design optimization, and more specifically relates to an optimization process modeling method supporting response surface estimation and updating. Background technique [0002] Multidisciplinary design optimization technology can effectively deal with the optimization problems of large-scale complex systems. However, the problem of high computational cost cannot be avoided in the specific multidisciplinary design optimization process, especially in the case of expensive simulation analysis involving FEA and CFD calculations. Will consume a lot of computing time. To cope with this difficulty, response surface technology was introduced into multidisciplinary design optimization. Based on response surface technology, a response surface that can accurately describe the input-output relationship in the original model was constructed to replace the original valuation model. The value replaces...

AI Technical Summary

Problems solved by technology

[0003] However, in the existing commonly used multidisciplinary design optimization software, such as ModelCenter, etc., the response surface function is used to construct the response surface when modeling the optimization process. During the operation of the optimization process, the response surface function is only used as a separate response surface The evaluation function can only statically reflect the mapping relationship generated by the response surface construction results. Once the response surface is generated, it cannot be changed. Therefore, the dynamic update of the response surface in iterative calculation cannot be realized during the process running.

Due to the lack of update function of response surface components, if you want to build the latest response surface, multiple response surface components will appear as the iteration progresses, making the calculation process very complicated, and iterative optimization using response surface serialization cannot be realized calculate

Images