Research On Characteristic Parameter Identification Method Based On DIC Measuring Information

With the development of image acquisition equipment,digital processing equipment and relative algorithms,digital image correlation technology has made it possible to measure structural deformation information in the full-field.This also provides sufficient data support for the calculation of model parameter identification by using deformation measurement information.However,since digital image correlation is a full field measurement,the amount of data obtained is very large.In recent years,extracting useful information and removing redundant noise from huge amounts of data have become important research topics.Therefore,it is necessary to study a method to solve the problem of a large amount of DIC data and difficulty in selecting points for digital image correlation measurement.In view of the technical difficulties above,the main work is as follows:1.The application fields of digital image correlation measurement and the relevant methods of characteristic parameter identification are introduced.On the basis of reviewing the typical methods,the principal component analysis method is proposed to compress the deformation data of structure surface.On the one hand,the application cost of huge data measured by digital image correlation technology is reduced.On the other hand,it achieves the goal of retaining the main features in structural surface deformation information.Furthermore,a structural parameter identification method based on DIC data principal component analysis compression is proposed to solve the problem of large amount of data and difficulty in selecting points.This method optimizes the difference between the DIC data PC A compressed and the finite element displacement data projected by the base vector as the optimal goal.The characteristic parameter model is established by least square method,and the Guass-Newton method is used to solve the model.2.The reliability of this method is verified by numerical simulation and test.Firstly,the inverse calculation of mechanical model parameters using compressed data is studied.It is proved under the framework of least squares method that the target residuals of the inverse model established using compressed data approach the inverse residuals of the original data.Secondly,the optimal material parameters,objective function and iteration times obtained by the direct inverse method and the method in this paper are compared.Finally,two examples are given to illustrate the effect of data compression on the inverse calculation of model parameters from the aspects of calculation accuracy,convergence speed and anti-noise.The results show that the proposed method can effectively improve the convergence speed of the inverse calculation of mechanical model parameters under the premise of significantly reducing the amount of data used,especially for the inverse calculation of multiple material parameters,and has higher accuracy and better stability.In dealing with noise,the algorithm shows faster convergence speed,stronger stability and noise suppression ability.3.Aiming at the non-linear problems in engineering practice,the proposed method is improved.It mainly includes two aspects.The first aspect of work is using the kernel principal component analysis instead of principal component analysis to compress the structural deformation information measured by digital image correlation.The other aspect is using the superlinear convergence NL2SOL instead of Guass-Newton algorithm to solve the mechanical model.The experimental results show that in the non-linear model,the objective function of the improved algorithm converges faster,and the obtained material parameters have higher accuracy and better stability.