The Order Identification Of Dynamic System Model And Temperature Effect On Modal Parameters

As developing of theory and application on damage identification based on dynamics theory, the research on damage identification is going deep day after day, but it's still suffering some problems on theory and application. This paper discusses these three aspects: The order identification of dynamic system model, temperature effect on modal parameters and parameter estimation based on Bayesian method.Because of time domain method's advantages that frequency domain method haven't, the identification theory based on time domain has become the emphasis in recent research. Some new theories'appearing such as the ERA,SIM and SSI give the time domain identification brightly foreground. But these theories need to determine the order of the model. But that is a difficult problem in research. This paper presents a new method, called residual-difference-expectation method, for determining the order of the state-space model of dynamic structural system. Impulse-response functions are employed to construct the Hankel matrices. The Hankel matrix can be decomposed as a H 1 matrix and a H 2 matrix by the singular value decomposition. H 1 and H2 comprises primary response signals and noise respectively. It's supposed that the signals of a dynamic system is mixed with the white noise only, then making use of the characteristic of zero-average-value of white noise, the order of a state-space model can be determined correctly. If the size of Hankel matrix has a proper value, though the dynamic system mixed with strongly noise, the method can identify the model's order exactly. Numerical examples in which response signals of a dynamic system are mixed with strong noise and actual signals measured form an experimental investigation on a reinforced concrete frame structure are studied.As going deep into the health monitoring and damage identification, it's found out that the effect of environment can't be ignored. In order to open out the temperature effect on modal parameter, impulse tests lasting 10 hours on a frame outside had been made, and the environment temperature and both surfaces'temperature of the middle columne have been tracked record when testing. ERA and SIM programs are made by author to deal with the data achieved in test. This paper analyzes the temperature effect on modal in quality and studies the temperature effect on concrete Young's modules. The tests result indicated that modal frequency decreases linear as temperature increasing, but damping ratio don't show any rule. Starting from choosing the prior probability, the process of Bayesian estimation is introduced in this paper and validated by test data. Because author's study object is the temperature effect on modal parameter, the data observed in test and the data after eliminating temperature effect have been compared, the result indicated that the precision of parameter estimation will be increased remarkably after temperature effect being eliminated. But the precision of Bayesian estimation increased unconspicuous than any else theory, the reason might be the prior probability and the quantity of swath.