基于支持向量机的故障诊断方法研究

A Study of Fault Diagnosis Method Based on SVM

针对故障诊断中存在的故障样本不完备问题,提出一种基于支持向量数据描述(SVDD)与支持向量机(SVM)相结合的故障诊断方法。该方法首先以正常状态下的数据样本与已知故障数据样本为整体建立数据描述模型、依据已知故障数据样本建立支持向量分类机模型,然后对输入的测试数据样本采用SVDD进行拒绝与接受处理,被接受的样本再利用支持向量分类机进行具体类别诊断;被拒绝的样本则为未知故障类型。数值试验表明,该方法可以有效处理故障样本不完备的故障诊断问题,能够对已知故障类型进行准确判断,并对未知故障类型给出提示,具有一定的实践意义。

Aiming at the fault samples incomplete problems existing in the fault diagnosis, we put forward a kind of fault diagnosis method based on Support vector data description (SVDD) and Support vector machine (SVM). This method ifrstly set up normal data samples with known fault data samples for whole data description model, on the basis of known fault data samples based support vector classiifcation machine model, based on a known fault data samples to build support vector machine model, Then the test data of the input sample SVDD is adopted to deal with the refused or accept,acceptable sample using the SVM to classify speciifc diagnosis; The rejected samples are unknown fault type. Numerical experiments show that this method can efifciently solve the fault diagnosis problem of incomplete fault samples, while the unknown fault type giving prompt better implementation of known fault types of speciifc judgments, which has certain practical signiifcance.