首页|特征向量预测的POD-Kriging模型

特征向量预测的POD-Kriging模型

扫码查看
特征向量在实际问题中有重要的含义,但实际问题常常所对应的矩阵维数是相当大的,直接求解特征向量是费时的.通过构建POD-Kriging模型对时变问题任意时刻各阶特征向量可以进行预测.对于时变系统求解特征向量可选取某些时刻的各阶特征向量样本矩阵,基于已知的各时刻各阶特征向量,通过POD方法预测各阶向量对应的一组基向量,由基的线性组合来预测任意时刻各阶特征向量,基系数则通过Kriging方法得到,从而可以用模型预测任意时刻的各阶特征向量.在预测过程中基向量只需要求一次,因此可以简化工作,且预测精度高.
POD-Kriging Model for Eigenvector Prediction
Eigenvector has important meaning in practical problems,but the matrix dimension of practical problems is often quite large,and directly solving eigenvector is time-consuming.In this paper,POD-Kriging model is constructed to predict each order eigenvector at any time of time varying problem.For the time-varying system to solve the eigenvector,sample matrix at some time can be selected.Based on the known eigenvector of each order at each time,POD method is used to predict a group of basis vectors corresponding to each order vec-tor,and the linear combination of basis is used to predict each order eigenvector at any time.The basis coeffi-cient is obtained by Kriging method.Thus,the model can be used to predict each order eigenvector at any time.In the prediction process,the basis vector only needs to be calculated once,so the work can be simplified and the prediction accuracy is high.

POD modelKriging modeleigenvector

郭晓珍

展开 >

山西能源学院强基学院,山西晋中 030600

POD模型 Kriging模型 特征向量

2022年山西省高等学校教学改革创新项目山西能源学院2022重点一流课程

J20221258202202

2024

山西师范大学学报(自然科学版)
山西师范大学

山西师范大学学报(自然科学版)

影响因子:0.512
ISSN:1009-4490
年,卷(期):2024.38(3)