摘要
针对船舶轴系扭振故障小样本事件,基于小波包Shannon熵与二叉决策树支持向量机(DAGSVM)理论建立一种轴系扭振故障诊断模型。首先通过船舶轴系扭振实验平台提取轴系扭振四种模式信号;然后利用小波包变换提取Shannon熵值,作为故障输入特征向量;最后利用K-CV交叉验证法提升支持向量机,对故障进行建模识别。试验表明,此法具有较高识别率,为船舶轴系扭振故障诊断提供了一种有价值的在线诊断方法。
Against torsional fault of the case of small sample in ship, this article based on the theories of Shannon entropy of wavelet packet and SVM--binary tree of the multi-class classification, this article sets up a model of fault diagnosis of ship shaft of torsional vibration. Firstly, relying on the experimental platform of ship shaft of torsional vibration, extracting kinds of signals of shaft operation mode and then, using wavelet packet transform, picking up the energy value of Shannon entropy acting as the fea- ture vectors of fault input. Finally, using K-CV cross validation method to promote DAGSVM and making an identification about the failure.The study result shows that SVM has higher recognition rate, providing a valuable on-line diagnostic method for the torsional vibration fault diagnosis of ship shaft.
出处
《电子技术应用》
北大核心
2011年第7期141-143,147,共4页
Application of Electronic Technique
基金
浙江省自然科学基金资助项目(Y1080929)
宁波大学学科项目(szxl1065)