基于支持向量机的海洋平台结构损伤识别方法

基于支持向量机(Support Vector Machine,SVM)提出一种近海导管架平台结构损伤识别方法。该方法利用小波包变换处理平台甲板上的加速度传感器信号得到小波包系数,从而实现特征提取,然后对小波包系数进行降维处理,再将降维后的样本输入SVM进行训练,并利用验证集对训练后的SVM进行验证。将该方法应用于渤海某平台的损伤识别中,结果表明:采用标准差较大的前5个或更多的小波包节点能量占比作为特征参数进行训练,训练后的SVM算法对验证集具有较好的识别效果。

Damage Identification Method for Offshore Platform Based on Support Vector Machine

A damage identification method for the offshore jacket platform based on Support Vector Machine (SVM) is proposed. The wavelet packet transform is performed on the acceleration sensor signal on the platform deck to obtain wavelet packet coefficients so as to realize feature extraction. The wavelet packet coefficients are dimensionally reduced in some ways. The dimensionally reduced samples are input into the SVM for training, and the trained SVM is verified by the test set. The method is applied to the damage identification of a platform in Bohai. The results show that the trained SVM algorithm has good performance for the test set having 5 or more wavelet packet nodes energy proportion with larger standard deviation as feature parameters.