【作者】 张周锁； 侯照文； 孙闯； 何正嘉；

【Author】 Zhousuo Zhang,Zhao Wen Hou,Chuang Sun,Zhengjia He (Xi’an Jiaotong Unversity,Xi’an,710049)

【机构】 西安交通大学机械工程学院；

【摘要】 针对现有混合智能故障诊断模型缺乏通用方法和混合框架,未能实现不同智能诊断方法的实质性融合和优势互补的问题,本文提出并构建了一种基于粒计算的混合智能故障诊断模型。该模型的核心是在邻域粗糙集中求取不同的邻域值,对故障特征属性集进行分层粒化,在不同粒度下获得核属性集;利用核属性集在相应粒度下构建神经网络和支持向量机两种子分类器;对所有粒度下全部子分类器的诊断结果通过评估矩阵算法进行融合集成。将该模型应用于株州高速机车轮对轴承的故障分类,结果表明分类精度随着粒度层的增加而不断提高,集成后的分类精度高于不同粒度下所有子分类器,体现了粒化分层的优势和不同智能诊断方法的优势互补。更多还原

【Abstract】 To solve the problem of lacking hybrid modes and common algorithms in hybrid intelligent diagnosis,this paper presents a new model of hybrid intelligent fault diagnosis based on granular computing.The core thoughts of the new model are granularity and combination.In the model,first the extracted features set is granularized in different levels by the reduction algorithm based on Neighborhood Rough Set.And then,the granularized core features are used to tain nural network and support vector machines as sub-classifiers in corresponding levels.Finally,the results of sub-classifiers in different granular levels are combined by criterion matrix algorithm as output of hybrid intelligent diagnosis.The model is applied to fault diagnosis in roller bearings of high-speed locomotive.The application results demonstrate that the classification accuracy is raised with the adding of granular levels,and the accuracy of hybrid results is higher than the classification accuracy of any sub-classifier.The proposed model exhibits the effect of granulation and superiority complement among different intelligent methods.更多还原