Incremental Kriging based sequential optimization experiment design method

Abstract

The invention discloses an incremental Kriging based sequential optimization experiment design method. The method comprises the following steps: 1, selecting initial experiment design; 2, initially building a model and verifying the model; 3, building a model of incremental Kriging (IKM); 4, optimizing a sampling rule; 5, updating the rule; and 6, building a model through DACE (design and analysis of computer experiment). The Kriging method used for building the model is effective, but the time of building the model through the Kriging abruptly increases as sampling points increase (when the quantity of the sampling points is more than 600 in case of two dimension). To solve the problem, an improved Kriging method is provided, and the improved Kriging method has the advantage that the time consumption can be minimized in case of a large amount of sampling points; due to the reduction of the model building time, the incremental Kriging based sequential optimization experiment design method can be widely applied to engineering simulation.

Description

technical field [0001] The invention relates to a sequence optimization design method to improve modeling efficiency in engineering applications. Background technique [0002] (1).Modern electromechanical products are becoming more and more complex, and modeling analysis (especially simulation model estimation) for them requires a lot of computing time. Although the computing speed of computers continues to increase, more and more complex simulation models, such as finite element analysis, computational fluid dynamics and other computationally intensive simulation models require a lot of computing time, so they cannot meet the needs of the industry for simulation analysis. To deal with the great challenge, the Kriging method has emerged and been widely used in the industry in the past few decades. This method can minimize the number of simulations of the source model in the optimization iteration process without affecting the accuracy of the simulation model, thereby reduci...

AI Technical Summary

Problems solved by technology

Furthermore, although one-time experimental design is very intuitive, it is inappropriate to use it to construct the Kriging model

First, for a large number of simulated estimates, one-time design of experiments is very time-consuming; second, it may be able to lead to improper use of the model when sampling a large number of data points

Images