Multidimensional Damage Identification Based on Phase Space Warping: An Experimental Study

A multidimensional damage identification scheme developed in previous work is modified and investigated experimentally. An experimental apparatus consists of a driven two-well magneto-elastic oscillator, where a cantilever beam vibrates in a magnetic potential field perturbed by two electromagnets. These electromagnets are activated by a computer controlled power supply and their terminal voltages are considered a two-dimensional damage variable. The effect of total change in the supply voltage of the electromagnets is approximately 4% shift in the experimentally measured natural frequencies of small oscillations in each well of the potential. Experimental runs are started in a nominally chaotic regime. The battery voltages are altered on specific trajectories in the damage (voltage) phase space. Damage identification is accomplished based on the elastic vibration data collected using laser vibrometers and a accelerometer. The phase space warping based damage tracking feature vectors are estimated using a new phase space partitioning scheme. The damage identification is achieved by applying smooth orthogonal decomposition to the obtained statistics. The effect of the data record size on the quality of reconstructed damage trajectory is investigated in a series of experiments. It is also demonstrated that the new partitioning scheme considerably improves signal-to-noise ration of the identified damage states.     

Damage detection in structures using modified back-propagation neural networks

A nonparametric structural damage detection methodology based on neural networks method is presented for health monitoring of structure-unknown systems. In this approach appropriate neural networks are trained by use of the modal test data from a ‘healthy’ structure. The trained networks which are subsequently fed with vibration measurements from the same structure in different stages have the capability of recognizing the location and the content of structural damage and thereby can monitor the health of the structure. A modified back-propagation neural network is proposed to solve the two practical problems encountered by the traditional back-propagation method, i.e., slow learning progress and convergence to a false local minimum. Various training algorithms, types of the input layer and numbers of the nodes in the input layer are considered. Numerical example results from a 5-degree-of-freedom spring-mass structure and analyses on the experimental data of an actual 5-storey-steel-frame demonstrate that neural-networks-based method is a robust procedure and a practical tool for the detection of structural damage, and that the modified back-propagation algorithm could improve the computational efficiency as well as the accuracy of detection Project supported by the National Natural Science Foundation of China (No. 59908003) and the Natural Science Foundation of Hubei Province (No. 99J035).