基于小波包样本熵和支持向量机的框架结构损伤识别

基于小波包样本熵和支持向量机原理,研究了钢框架结构的损伤定位识别方法.分析在冲击载荷作用下框架结构的动力响应,对加速度信号进行小波包分解,建立小波包样本熵的损伤指标,采用支持向量机原理,识别结构损伤位置以及损伤程度.研究表明,该方法能够利用单一的传感器,实现理想的识别效果,且具有一定的适用性和鲁棒性,在60 dB的噪声水平环境中损伤定位识别结果在90％以上,在40 dB的噪声水平环境中,损伤程度识别结果在90％以上,框架实验模型研究表明,柱的损伤识别精度要高于梁的损伤识别精度.

Damage identification of frame structure based on wavelet packet entropy and support vector machine

The combination of wavelet packet entropy and support vector machine is used for identify the damage condition of frame structure. The acceleration response under the impact load is decomposed into wavelet packet components, and then the damage index of wavelet packet sample entropy is computed, which is used to identify the structure damage position and the degree by support vector machine. The research shows that this method can realize the ideal recognition effect. It has applicability and robustness. The proposed algorithm is sensitive to the damage and has high recognition accuracy. In the noise level environment of 60dB, the result of damage location identification is more than 90%, and in the noise level environment of 40dB, the result of damage degree identification is more than 90%. The experimental model study of the frame shows that the damage identification accuracy of the column is higher than that of the beam.