基于VMD-FCM-RLSSVM的多模过程故障诊断方法

摘要

In order to ensure normal process of the industrial process,it is needed to identify the fault in time to realize fault diagnosis.However,traditional methods are based on single mode,which is single mode,and so it is not suitable for multimode process.To solve this problem,a combined fault diagnosis method is proposed by variational mode decomposition (VMD),fuzzy C means (FCM) and recursive least squares support vector machine (RLSSVM).Firstly,VMD method is adopted to denoise the collected data,then,FCM is applied to carry out the modal discrimination.Finally,the fault diagnosis is realized by using RLSSVM method.The simulation results of CSTH process show that the proposed method can diagnose the fault rapidly and effectively.%为了保证工业过程的正常进行,需要及时地辨识出故障以实现故障诊断.然而,传统的故障诊断方法多是基于单模过程而不适应于多模态过程.为了解决这个问题,本文提出了一种变分模态分解(VMD)、模糊C均值(FCM)及递推最小二乘支持向量机(RLSSVM)相结合的集合型故障诊断方法.本文首先介绍了VMD方法对所采集的数据进行去噪处理,然后利用FCM进行模态区分,最后利用RLSSVM方法实现故障诊断.通过CSTH过程的仿真结果表明,该方法提高了诊断效率和性能,能够快速、有效地诊断出故障.