Modal Parameter Identification Of Large-scale Structures Based On Subarea Measurements

Modal analysis technique is widely used in such fields as automobile, aerospace, ship‐building and bridge industries. Although the theory algorithm has been very mature, when applied to many large complex engineering structures, many deficiencies still exist in practical application, one of the biggest problems is to get the overall structural response signals which are satisfying and with high signal‐to‐noise ratio through artificial incentive method. For the limitations in large complex structure application, the thesis studies them from two aspects. On one hand, eliminate or reduce the noise in measured signals by Singular Value Decomposition SVD technique. On the second hand, obtain input and output signals through subarea measurements, then carry out modal parameter identification for every subarea, and integrate all subarea mode shapes into the whole structure mode shape.The random noise can be effectively reduced by using SVD technique. First construct Hankel matrices with clean simulation signals and contaminated simulation signals, then execute SVD for the Hankel matrices, compare the singular values and analyze the effects of noise to singular values. For the most important process, determining the effective number of singular values, the thesis analyzes three methods, which are singular value curve, increment of the singularity entropy and curvature spectra of singular values, then point out the deficiencies. According to the relationship between the extreme value point number of the noise elimination signal and the selected number of singular values, a new method named extreme value point number is proposed, simulation and experiment signals validate the new method.When doing experimental modal analysis for large complex structures, the modal parameter identification based on subarea measurements can solve the problem of getting satisfying response data. For the original method, the thesis presents the existed deficiencies, and put forward a new method, with the original subareas added with ’transition areas’, to divide the subareas on the whole structure. A new concept called ’Shape Reliability’ and its computing method are proposed refer to the reference points in the transition area, and make it possible to forecast the outcome of integrated modal shape. According to a mathematical equation, the whole structure shape integration method is brought out. With this method, subarea shapes are integrated into a whole structure shape after the subarea modal parameters identified. The thesis compares the modal parameter identification methods based on subarea measurements and overall structure measurements, using small simple structure and large complex structure respectively, and through computing the errors between two methods, the thesis validates and analyzes the relationship between errors and ’Shape Reliability’. The orthogonality of subarea shapes are affected for transducers are relatively less. According to the rules of linear independence and condition numbers, reduce the number of transducers by gradual reduction method, and compare the two methods. Based on the assumption of transducers arrangement on subarea, the thesis analyzes the effects of subarea to the linear independence of modes.