A Reliable Suset DIC Method For Deformation Measurements Of Similar Models And Its Application

The digital image correlation(DIC)method is an important non-contact optical metrology,easy to operation,and applicable for the high measurement accuracy and the full-field deformation measurement.The high accuracy deformation measurement of the DIC method needs to meet two basic conditions.One is the constant variation of intensity(the variation of intensity is constant),the other is the deformation continuity.However,speckles falling off,invalid areas and cracks are sometimes inevitably produced on the surfaces of similar models in similarity simulation experiments,which takes a great challenge for the high accuracy deformation measurement of the DIC method.Therefore,it is necessary to develop a novel DIC method which can be used to deal with instability speckle fields and discontinuous deformations.Several problems of the traditional DIC method for deformation measurements of similar material models were studied in this thesis,including strain measurements,deformation initial value guesses,speckle fields failures and discontinuous deformations,and a reliable subset DIC method was proposed based on image feature matching.The least absolute deviation fitting method was introduced to obtain reliable strains for non-uniform deformations and displacement fields with outliers.As the least absolute deviation fitting problem cannot be solved by analytical methods,a hybrid particle swarm optimization algorithm based on simulated annealing was used.The accuracy and correctness of the least absolute deviation fitting method for strain measurements were verified by use of a numerical experiment on the formation of a fictitious shear band and a physical experiment on a fault slip.The deformation initial value guess method based on SIFT(scale invariant feature transform)feature matching is an efficient method.However,some false matching points in SIFT feature matching have a serious effect on the accuracy and effectiveness of obtained initial values.The great errors of the initial values caused by false matching can be eliminated through introducing the least absolute deviation fitting method in the initial guess method based on SIFT feature matching.The accuracy of the new initial guess method was verified by use of numerical experiments on the translations of a speckle pattern and the formation of a fictitious shear band,and limitations of the new initial guess method were verified by use of a physical experiment on a fault slip.A reliable subset DIC method was proposed based on SIFT feature matching,the DIC method and the least absolute deviation fitting method.In the proposed method,firstly,reliable subsets were selected by feature matching and displacement initial values of reliable subsets were obtained;then,deformation parameters of reliable subsets were obtained by correlation operation;finally,deformation parameters of monitored points were determined by the least absolute deviation fitting of displacements of reliable subset center points in certain areas around monitored points.The effectiveness of the proposed method was verified using a physical experiment on a fault slip.The reliable-subset DIC method was improved using the classification of center points of reliable subsets in fitting regions and the weighting of correlation coefficients for the displacement field measurements of cracks.The effectiveness of the improved method was verified using numerical experiments on the formation of horizontal and inclined mode I cracks and physical experiments of similar material specimens in uniaxial compression.The improved method was further verified for both instability speckle fields and discontinuous deformations by use of physical experiments on the proportion of similar materials.There are 95 figures,10 tables and 172 references in this paper.