Application Research On The Structural Health Monitoring Based On Hilbert-Huang Transform

Structural health monitoring technology is an active research issue both in professional and academic area. Vibration signal processing and analysis is one of the important means in structural health monitoring and damage detection. After reviewing some Time-Frequency methods and analyzing their characteristics and disadvantages, Hilbert-Huang Transform (HHT) is chosen as the research object in this thesis. End-effect issue and approach to deal with end-effect, modal parameter identification and damage detection based on HHT are investigated.Hilbert-Huang Transform is a new technology for processing nonlinear and non-stationary signal. This method consists of two successive parts, Empirical Mode Decomposition (EMD) and Hilbert Spectral analysis (HSA). Any arbitrary nonlinear or non-stationary signal can be decomposed into a number of intrinsic mode functions (IMFs) by EMD. Then Hilbert spectrum analysis is performed on each IMF, and Hilbert spectrum of the corresponding IMF is obtained. Hilbert spectrum of all IMFs are summarized to get Hilbert spectrum of the original signal. Hilbert spectrum obtained by this way describes the original signal in joint time-frequency domain, and possesses high time-frequency resolution. The disadvantages existed in several signal processing methods based on Fourier analysis are totally overcomed. However, there is a troublesome end-effect issue due to using spline interpolation to get the upper and lower extreme envelopes. Mirror extension, radial basis function (RBF) neural network prediction and their combination are investigated to restrain end-effect in EMD. And a simulated signal and an acceleration record from shaking table test of a 12-stroey reinforced concrete frame model are processed by these methods and further decomposed by EMD. The results indicate that combination of these two methods can improve the decomposing results effectively, especially for low-frequency IMF components.According to basic theory of modal parameter identification, combination of EMD and Random Decrement Technique (RDT) is adopted to extract the single-mode free decay response. And after obtaining instantaneous amplitude and instantaneous phase by Hilbert transform, natural frequencies and damping ratios can be identified. Then acceleration records of shaking table test on a 12-storey reinforced concrete frame model are processed and modal parameters are identified.Because of the defect in processing time series data, traditional structural damage identification research based on vibration mostly assumes that the system is linear before and after damage. Damage is detected by comparing dynamic parameters of the structure. In this thesis, HHT is adopted to process shaking table test data of the 12-storey reinforced concrete frame model. Identified instantaneous characteristics of the model structure by processing time domain signal, namely instantaneous frequency, can reflect the time into nonlinear state and the level of nonlinearity. Then HHT time-frequency spectrum and instantaneous energy spectrum are adopted to identify the time when the damage starts and HHT relative marginal spectrum is adopted to identify damage location.