Identification Of Distributed Dynamic Loads And Its Application To Buffet Load Prediction

In order to ensure the reliability and safety of the designed engineering structures,the estimation of dynamic load environment is a very important and necessary task.However,in most engineering practical cases,it is impractical to meas ure the dynamic loads directly.Under this circumstance,it would be beneficial if the dynamic loads can be estimated from the measured dynamic response of the vibrating structure,which are relatively easy to be acquired.The study on the identification of dynamic load has not only the academic significance but also engineering application merits.This thesis devotes to study the identification of the distributed random dynamic load and its application to buffet load predication.The study of this thesis has four aims:(1)identify the multi-point random dynamic load and analyze the error of the random dynamic loads;(2)estimate the random dynamic load based on blind source identification;(3)determine the distributed random dynamic load acting on the linear elastic structures;(4)study the inverse estimation method for the buffet load.The main research work and achievements in this thesis are as follows:(1)The basic principle of the dynamic load identification is introduced,which includes the establishment of dynamic load identification model?acquisition of dynamic responses?the selection of dynamic load identification method and regularization process.The difficulties in every part are summarized and specfic methods are given to solve these problems.Additionally,the paper introduces some basic knowledge,which is used in the course of load identification,and the numerical examples are given to illustrate the calculation of PSD(Power Spectral Density,PSD)of the random signals and the application of the regularization toolbox in Matlab.(2)A random dynamic load identification model considering both frequency response function error and responses error is established,the upper bound of relative error of random dynamic load identification is also derived,and the error propagation and amplification are analyzed.In order to improve the accuracy of random dynamic load identification,two methods(i.e.the Tikhonv regularization method based on weighted matrix and total least squares method based on weighted matrix)are proposed.Additionally,the selection principle and the specific form of the weighting matrix are given and validated analytically and experimentally.The experimental results show that the two proposed method are superior to the unweighted method.(3)In order to deal with the problem of frequency response function matrix not to be measured by traditional method,a new method is proposed to identify the random dynamic load by only dynamic response.In this method,the second order blind source identification is applied to identify the modal parameters(i.e.natural frequency?damping ratio?mode shape)firstly,and then the modal loads are identified by the modal parameters and dynamic responses in the modal space,at last,the random dynamic loads are identified in the physical space.The experimental results indicate that the proposed method can identify the random dynamic loads acting on the model with good accuracy.(4)A frequency domain approach is extended to identify the distributed random dynamic loads acting on the linear elastic structures.In this method,the distributed random dynamic loads can be transformed into a linear combination of time coefficients and orthogonal functions.Under the modal space,the identification problems of distributed random dynamic loads can be solved by estimating the time coefficients.These time coefficients can be calculated by the orthogonal functions.In order to deal with the ill-posed problems,the POD method(Proper Orthogonal Decomposition POD)and the Tikhonov method are introduced,and the regularization parameters are determined by GCV(Generalized Cross Validation,GCV)method.The results indicated that the proposed method can identify the distributed random dynamic loads with good accuracy.(5)In order to identify the buffet loads acting on the vertical tail of aircraft,a new frequency method is proposed to identify buffet loads.Firstly,the buffet load can be transformed into a linear combination of time coefficients and orthogonal functions.Thus the buffet load can be identified by estimating the time coefficients.In order to address the ill-posed problem encounted in the identification process,the Tikhonov method are introduced,and the regularization parameters are determined by GCV method.To validate the proposed method,a vertical tail model is built.The dynamic loads,which simulate the buffet load,are calculated by CFD(Computational Fluid Dynamic,CFD)software,and these loads are exerted on the 25 regions of the surface of the vertical tail to calculate the acceleration responses.Finally,these acceleration responses are adopted to estimate the buffet loads and the identified results indicated that the proposed method can identify the buffet loads with good accuracy.