源域多样本集成GFK的不同工况下滚动轴承寿命状态识别

Life status identification of rolling bearing under different working condition based on multi-source integrated GFK

针对不同工况下滚动轴承寿命状态识别时训练样本与测试样本分布差异导致寿命状态无法有效识别的问题,提出基于源域多样本集成(Geodesic Flow Kernel,GFK)的滚动轴承寿命状态识别方法。首先,采用无重复均匀随机抽样对源域类间样本进行多次等量随机抽样得到源域内部多个训练样本以充分挖掘源域样本信息;其次,将源域内部多个训练样本和目标域测试样本输入GFK,分别计算每个源域训练样本与目标域测试样本的测地线核矩阵以充分利用源域样本信息并提升GFK迁移学习能力;最后,利用核矩阵构造核分类器并输出分类结果,采用一致性投票对所有源域训练样本下目标域测试样本的分类结果进行集成以提升目标域测试样本的识别准确率。不同工况下滚动轴承寿命状态识别实验验证了所提方法的可行性和有效性。

Aiming to solve the problem that the life state is not effectively identified due to the distribution difference between training samples and test samples under different working conditions,a life status identification method based on source-domain multi-sample integrated GFK is proposed.First,uniform random sampling without repetition is used in the source domain to get multiple training samples from the source domain to fully excavate the source sample information.Secondly,the multiple training samples in the source domain and the test samples in the target domain are inputted into GFK and the geodesic kernel matrices of each training sample in source domain and test sample in target domain are calculated respectively to make full use of source domain sample information and improve GFK migration learning ability.Finally,the kernel classifier is constructed by kernel matrix and the classification result is output.The classification result of the test samples in target domain under all training samples in source domain is integrated by using the consistency vote to improve the recognition accuracy of the target domain test samples.The experimental results of rolling bearing life state under different working conditions verifies the feasibility and effectiveness of the proposed method.