Study On Identification Method For Structural Parameters With Incomplete Information

With the rapid development of society, more and more people have devoted their attention to safety. The structures of industry and civil engineering are the primary place of human to live and work in. So it's parameter identification and health monitoring is becoming an interesting focus and is studied by many researchers. Although many parameter identification methods have been put forward and made great success in mechanical engineering, the successful application in industry and civil engineering is very few. The mean reason is that structures in industry and civil engineering are much more great and complex. The traditional identification methods in mechanical engineering are based on the assumption that the input and output information are all known, which is difficult to be realized in industry and civil engineering. One side, because of their huge volumes, forced excitation is often impossible. On the other side, the structures are always vibrated by ambient excitation such as ground motion. But, due to their randomness, it is hardly to be measured accurately. In addition, as the feature of industry and civil engineering, their structure will have many degrees of freedom. To measure all of their responses is impossible either. Hence, how to identify structure with incomplete information is an important problem in this field. This paper is devoted in the study of this problem.This paper can be divided into six chapters. In the first chapter, the importance of parameter identification and health monitoring for structures is introduced. And many methods of this filed in recent years are reviewed and summarized. In the second chapter, the basic theory of inverse problem for structure identification is discussed, and incomplete information in dynamic system is put forward. To solve the ill-posedness raised by incomplete information, two basic ideas are presented. The first idea is to identify the structure by eliminating the unknown input information. And the second idea is to identify the structure by modifying unknown input information in the iterative process.The third chapter is the emphases of the paper. In this chapter, the identification method for linear system in time domain is studied. By using the first idea, a two-step method for structure with partly unknown input information is put forward. And by using the same idea, another method for shear structure with unknown proportional input information is yielded. In addition, as to underground structure, by introducing the virtualparameters, the unknown restriction caused by around soil is predigested. Based on this, the identification method of multi-parameters for underground structure is advanced. As to the second idea, many research papers have been published. Based on the study of theory and numerical examples, this paper point out the iterative method for structure with partly unknown input information has applied condition. When the applied condition is absent, the correct result can't be obtained. Many paper also discussed the iterative methods for structure when the unknown input information is proportional, such as ground motion and wind load. By comparing and discussing the former methods, this paper advanced a new method called the weighted average method. Based on this new method, this paper does the further discussion and studies the identification approach which can be used to the situation when multiple input information is unknown.The fourth chapter studies the identification method for nonlinear parameter system in time domain. By introducing the "eliminating" idea, an approach to identify common nonlinear parameter system with incomplete input information is presented. At the same time, a new damp model is introduced and developed. As to the structure with this new model, an iterative process based on the least square criterion is put forward. Because this process accords with the transfer relation of error in the "modifying" idea, this process can be combined with the "modifying" idea. So the structure with this new model can be estimated when the unknown input information is proportional.The fifth chapter studies the identification method in frequency domain. Firstly, this chapter discusses the iterative process based on the least square criterion using the complex mode parameter. Secondly, this chapter studies the optimization identification method using the limited transfer functions. As to the structure with separate mode parameter, this chapter builds up an objective function using mode shapes with limited measured points. Thirdly, by introducing the "eliminating" idea, this chapter builds an objective function without input information. Then the structure can be estimated when the input information is unknown.The sixth chapter summarizes the main content of this paper and predicts the future development of this field.