基于MVRVM回归和RVM二叉树分类的自确认气动执行器故障诊断算法

Self-Validating Pneumatic Actuator Fault Diagnosis Based on MVRVM Regression and RVM Binary Tree Classification

为了解决自确认气动执行器的故障诊断问题,提出了一种基于多变量关联向量机(MVRVM)回归和关联向量机二叉树分类的气动执行器故障诊断方法,该方法利用多变量关联向量机回归建立气动执行器的正常模型,然后将实际输出与模型输出比较,产生残差作为气动执行器的非线性故障特征向量。以残差作为输入建立关联向量机二叉树多分类机,诊断气动执行器故障类型。利用DABLib生成的故障数据对所研究方法进行了验证,并与基于RVM一对一分类的故障诊断方法进行了比较,结果表明该方法是解决气动执行器故障诊断的小样本和非线性问题的一种有效方法。

In order to solve the fault diagnosis problem of self-validating pneumatic actuators, a fault diagnosis approach based on multi-variable relevance vector machine （MVRVM）and relevance vector machine （RVM）binary tree classification is proposed. The MVRVM regression is used to establish the normal model of the pneumatic actuator. The residuals generated by comparing the output of the model and the actual actuator are used as the nonlinear features of the pneumatic actuator. Then the RVM multi-classifier based on binary tree is established and trained by the residuals, which is used to identify the condition and fault pattern of the actuator. The proposed approach is verified using fault data generated by DABLib model and compared with RVM one-against-one multi-classification method. The results indicate that the proposed approach is a valid method to resolve the small sample and nonlinear problem in pneumatic actuator fault diagnosis.