Research Of HHT In Structure Health Monitoring

In modern society, structural health monitoring system is widely used in aircraft, bridges, offshore platforms and other large and complex structures. The vibration based structural damaged detection technology plays a very important role in structural health monitoring system.In order to develop the vibration based structural damaged detection technology, several problems need to be solved, such as how to extract damage characteristic vectors, which are sensitive to damages, from structural dynamic responses; how to reduce the interferences from noises to damage characteristic vectors. The paper mainly focuses on how to construct and extract the parameter which is more sensitive to damages and anti-interference to noises. In order to achieve the object, A new method is proposed to improve the Hilbert-Huang Transform and Instantaneous frequency is used to improve its sensitivity for damage. It can help to forms a set of Hilbert-Huang Transform based online health monitoring method for large complex structures, which significantly improved the ability of damage detection.The main works are showed as follows:Firstly, principle of Hilbert-Huang Transform and causes of end effect in Empirical Mode Decomposition are analyzed, and a new method, end sifting method, is proposed to improve accuracy of Empirical Mode Decomposition, which avoid drawbacks of composing or forecasting end points in previous methods by judging and sifting every possible location of extreme points. The candidate number and end effect size are set to guarantee a most suitable Intrinsic Mode Function by controlling the results of judging and sifting. The end extending method is used to eliminate energy leakage problem of Hilbert Transform.Secondly, dynamic models of wing-box with intact and several kinds of damage levels are built to gain their dynamic responses through instantaneous analysis.Finally, Instantaneous Frequency of improved Hilbert-Huang Transform is taken as damage characteristic vector, and their relative variations from various damage levels to intact level are considered to show their sensitivity for damages. At last, the influence of noise is also discussed.The analyzed results show that Instantaneous Frequency is more sensitive to small damages and more anti-interference to noise with 1% level. The pattern classification function of artificial neural networks can help to build the mapping relation between damage characteristic vectors and damage statuses in various levels, and the work lays the foundation for online health monitoring of complex structure.