Study On Multi-fidelity Method Based On Compressed Sensing

In the research of many engineering problems,there are usually a lot of uncertain factors,so when applying the uncertainty quantification method,the generated expansion will suffer from the influence of dimension disaster.In order to deal with this influence,this paper combines the compression perception and uncertainty measurement to study the multi fidelity method.In the process of numerical simulation,random variables with uniform distribution are selected as sample points by using random sampling method,Legendre polynomials are used as the basis of polynomial chaos,and the multi fidelity method based on compression perception is studied by combining non embedded polynomial chaos.Single fidelity model and multi fidelity model are established on two-dimensional analytic function examples The validity of this method is proved by numerical simulation.