Abstract:In order to reduce the influence of edge-effect of continuous wavelet transform on the identification of modal parameters, artificial neural network is introduced to extend free decay response signal in both directions. A multiple-input and single-output back propagation network is devised to predict the signal at the next time step using a finite number of current samples. The training samples are obtained by simple combination of the original signal. Both numerical and experimental studies are performed to demonstrate the proposed procedure and verify its practicability. The experimental system is a force sensor developed for minimally invasive surgery robot. The results show the method works very well in identifying damping ratios and undamped natural frequencies from short signal.
代煜;张建勋. 基于小波变换和反向传播网络的模态参数辨识[J]. , 2012, 31(3): 55-59.
DAI Yu;ZHANG Jian-xun. Identification of modal parameters based on wavelet transform and back propagation network. , 2012, 31(3): 55-59.