Structural Damage Identification Based On The Particle Filter And Reliability Analysis

As it is known to all, from the beginning of the structure getting into use, it faces the coupling effect of environmental erosion, material aging, the long-term effect of load, mutation & fatigue effect, and other factors, which will unavoidably trigger the resistance decay of the structure and damage accumulation, ultimately leading to structural damage and destruction. Investigation and study of the China Academy of Building Research on the building structures show that owing to the problems of design, construction and management of the building structure, varying degrees of damage exit in the majority of structures. On the other hand, many China’s early industrial and civil buildings, office buildings, as well as a number of bridges has been close to the design reference period. These structures have been subjected to different degrees of damage in the complex service environment. Therefore, the accurate judgment of the damage location and degree can not only guarantee the safety, integrity, and reliability evaluation of the structure, but also can make scientific and reasonable maintenance and reinforcement scheme, reduce maintenance costs, improve maintenance efficiency and the social security, prevent people’s property from unnecessary losses, enhance mutual understanding, and deepen the research on the performance of the structure. In addition, this will also has important practical significance in promoting the further development and research of the structural engineering.With more and more attention being paid to the structural health monitoring, promoting the reaserch on the methods of the structural parameters based on the excitation and response of the structure. Current structural damage identification is based on the identification of structural parameters, that is, when the engineering structure is damaged, the parameters of the structure must be changed. At this time if the change of structure parameters can be perfectly recognized, the structural damage characteristics can be discovered on time.Bayesian model correction method is the use of statistical Bayesian principle, which is embedded a set of possible probability models into a set of possible probability models, making the structural model quantify model predictions and observe the uncertainty possible. The basic idea of Bayesian identification method is to view the estimated structural parameters as random variables. Then observing and analyzing the parameters associated with other variables to infer the value of this parameter. For the linear Gauss state space model, the analytical expression of the posterior probability density can be obtained by the Caiman filtering method. The actual data is usually more complex, including non Gauss and non-linear situation. To solve this problem, many researchers have done a lot of research, putting forward the method of extending Calman filter and Gauss sum. But these two methods produced poor recognition results because they have not considered all the statistical characteristics in the process. And for the case of non Gauss nonlinear case, the problem will become more complex. As such, particle filter algorithm gets extensive attention from people because it can apply to the identification of linear and nonlinear structural parameters.In this paper, the author will make analysis in terms of the particle filter in the structural parameters identification and structural damage identification. The main research contents are as follows:1) This paper introduces the development of structural damage identification firstly, the main application field of particle filter, and the present situation of the research in details; then, it makes the inductive introduction of the advantage of particle filter in the field of structural parameter identification and points out that the outstanding advantages of the particle filter in the field of structural parameter identification; furthermore, it concludes some difficulties and problems that people encountered in the research on the application of particle filtering; finally, it illustrates the main content of this paper.2) This paper presents and introduces a method for identification of structural parameters based on maximum likelihood value. Under different levels of Gauss and non Gauss noise, it carries out the numerical simulation of structure of single degree of freedom and summarizes the advantages and disadvantages of the method of structural parameter identification based on maximum likelihood value.3) Introduces and explains the basic principles of particle filter algorithm on structural parameter identification and structural damage identification. At the same time, the single degree of freedom structure parameter identification process which is based on particle filter is described in detail and system. Under different levels of Gauss and non Gauss noise, it carries out the numerical simulation of structure of single degree of freedom and summarizes the advantages and disadvantages of the method of structural parameter identification based on particle filter. In addition, briefly analyzing the reliability of the results of numerical simulation based on the single degree of freedom structure parameters and damage identification.4) Based on the study of damage identification of structure of the single degree of freedom, the paper extends it to the multi degree of freedom. It also makes numerical simulation under different Gaussian and non Gaussian noise level. After analyzing the structural damage identification results, it summarizes the structural damage identification of multi degree of freedom based on the performance of particle filter. The paper also does the experimental analysis of the parameters and damage identification performance of the algorithm on the aluminum framework with four layers and compares the experimental recognition results and the structure of interlayer stiffness measurement value, proving the effectiveness of the particle filter algorithm in the stiffness damage identification of structure of multi degree of freedom.5) The paper concludes the advantages of particle filter in the field of structural damage identification and mentions the application prospect of the particle filter algorithm. As a new method in the field of structural health monitoring, the author summarized the existing problems and deficiencies based on their own research work on the particle filter algorithm. Last but not least, some discussions and prospects on the next step of particle filter research are discussed.