An improved principal component analysis method based on wavelet denoising preprocessing for modal parameter identification

Authors:

Haixiao Chi, Huaqiao University, Xiamen, China

Feng Hou, Huaqiao University, Xiamen, China

Zongwen Fan, Huaqiao University, Xiamen, China

Wangping Guo, Huaqiao University, Xiamen, China

Meizhen Chen, Huaqiao University, Xiamen, China

Abstract:

Purpose. Accurate identification of modal parameters is an important prerequisite for structural health monitoring and damage identification.

Methodology. Wavelet analysis is one of the most advantageous methods because it has the ability to represent the local features of the signal in time and frequency domain. The modal parameter identification effectively achieved using principal component analysis (PCA), can be regarded as a type of system recognition.

Findings. Because PCA is sensitive to Gaussian measurement noise, the authors propose a novel method that combines wavelet denoising with PCA. The technique was applied to modal parameter identification.

Originality. The signals are decomposed into wavelets with several layers, and the resulting wavelet coefficients are preprocessed according to a threshold. The signals are then reconstructed to reduce the effect of noise. The research on this aspect has not been found at present.

Practical value. Simulation results for beams show that the proposed method is able to recognize the main modal shapes and eigenfrequencies. Additionally, it can improve the precision of the identified modal parameters and extract some previously lost modes.

Список літератури / References:

1. Khanmirza, E., Khaji, N. and Majd, V.J., 2011. Model updating of multi-storey shear buildings for simultaneous identification of mass, stiffness and damping matrices using two different soft-computing methods. Expert Systems with Applications, Vol. 38, No. 5, pp. 5320–5329.

2. Bakir, P.G., Eksioglu, E.M. and Alkan, S., 2012. Reliability analysis of the complex mode indicator function and Hilbert transform techniques for operational modal analysis. Expert Systems with Applications, Vol. 39, No. 18, pp. 13289–13294.

3. Wang, J., Barreto, A., Wang, L., Chen, Y., Rishe, N., Andrian, J. and Adjouadi, M., 2010. Multilinear principal component analysis for face recognition with fewer features. Neurocomputing, Vol. 73, No. 10–12, pp. 1550–1555.

4. Akhtar, M.T., Mitsuhashi, W. and James, C.J., 2012. Employing spatially constrained ICA and wavelet denoising, for automatic removal of artefacts from multichannel EEG data. Signal Processing, Vol. 92, No. 2, pp. 401–416.

03_2016_Haixiao
2016-07-29  451.8 KB  745