提升小波包和BP-AdaBoost模型在大跨斜拉桥拉索损伤识别中的应用

摘要

In order to effectively recognize the damage in engineering structures, the cable damage identification methods for long-span cable-stayed bridges was proposed based on lifting wavelet packet feature extraction and BP-AdaBoost (Back Propagation neural network, the Adaptive Boosting) model. First of all, the vibration signal was decomposed using lifting wavelet packet (WP) analysis based on lifting frame. Then, the corresponding characteristic vector was established by the energy accumulating variation value of the lifting WP component energy. The vector was used to identify the damage location of the cable of the cable-stayed bridge. Finally, the BP- AdaBoost model was established. Combining AdaBoost algorithm with BP neural network, the damage level of the cable of the long-span cable-stayed bridge was identified. The influence of noise on the algorithm was also studied. The numerical results show that the proposed method can be used to effectively identify the cable damage of long-span cable-stayed bridges.%为了有效地进行工程结构的损伤识别，提出基于提升小波包特征提取和BP-AdaBoost模型的大跨斜拉桥拉索损伤识别方法。该方法首先利用提升框架，将结构损伤前后的振动测试信号进行提升小波包分解，提取小波包信号分量能量并将能量累积变异值作为特征值，识别斜拉索损伤位置，然后以此建立BP-AdaBoost ( Back Propagation neural network，Adaptive Boosting)模型，利用AdaBoost算法和BP神经网络相结合的方法对大跨斜拉桥拉索的损伤程度进行识别，并研究噪声对该算法的影响。数值分析结果表明，采用基于提升小波包和BP-AdaBoost模型相结合的方法能够有效地识别大跨斜拉桥拉索损伤。