基于在线改进符号序列熵与逻辑回归模型的岸桥起升减速箱在线退化评估  

Online Degradation Assessment of Shore Bridge Hoisting Gearbox Based on Improved Symbolic Sequence Entropy and Logistic Regression Model

作　　者：王微[1] 王冰 胡雄[1] 孙德建[1] WANG Wei;WANG Bing;HU Xiong;SUN Dejian(Logistics Engineering College,Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]上海海事大学物流工程学院,上海201306

出　　处：《上海交通大学学报》2021年第10期1272-1280,共9页Journal of Shanghai Jiaotong University

基　　金：国家高技术研究发展计划(2013AA041106);国家自然科学基金(31300783);中国博士后科学基金(2014M561458)资助项目。

摘　　要：针对岸桥起升减速箱的退化状态评估问题,提出一种基于在线改进符号序列熵(O_ISSE)与逻辑回归的退化状态在线评估方法.首先引入阈值因子,保留信号变化方向与幅值的“粗粒化”信息,降低原算法对于冲击成分的“敏感性”,提出改进后的符号序列熵(ISSE).采用滑动窗Weibull拟合的方法有效滤除ISSE特征序列中的波动影响,形成O_ISSE.最后,训练并建立逻辑回归模型,在线计算得到未知样本的健康因子H值,实现未知样本的状态识别.采用上港某码头在线监测的起升机构减速箱全寿命数据进行实例分析.研究结果表明,所提ISSE和逻辑回归模型方法能够挖掘信号中的复杂度变化规律,准确地跟踪并识别性能退化状态.Aimed at the problem of degradation assessment of shore bridge hoisting gearbox,an online evaluation method of degradation state based on improved symbol sequence entropy(O_ISSE)and logistic regression is proposed.First,a threshold factor is introduced to retain the“coarse-grained”information of the signal change direction and amplitude,reduce the“sensitivity”of original algorithm to the impact component,and propose an improved symbol sequence entropy(ISSE).Then,the sliding window Weibull fitting method is used to effectively filter out the influence of fluctuations in the ISSE characteristic sequence to form O_ISSE.Finally,a logistic regression model is trained and established,and the H value of health factor of the unknown sample are calculated to realize its status recognition online.The example analysis of the life data of the hoisting gearbox of a dock in Shanghai is conducted.The results show that the ISSE and logistic regression model can describe the complexity of signal,track and identify performance degradation status accurately.

关 键 词：岸桥 退化特征提取 符号序列熵 WEIBULL分布 减速箱 

分 类 号：TH133[机械工程—机械制造及自动化]