论文部分内容阅读
针对基于模型的故障诊断流程中故障检测和故障识别两个关键问题,提出了一种基于神经网络的实现方法.首先利用BP神经网络进行参数估计,并结合系统模型进行故障检测;然后采用ART2神经网络进行数据聚类,并基于聚类结果进行系统故障识别;最后,设计实现了基于BP/ART2神经网络的故障诊断系统.基于BP神经网络的参数估计方法可以准确地估计诊断对象在不同状态下的参数,为故障检测提供有效依据;基于ART2神经网络的数据聚类不仅可以识别对象的已知故障类型,还可以识别出未知故障,对先验信息较少的系统进行故障识别更具有效性.通过永磁直流电机故障诊断案例的应用,证明方法能具有一定的工程实用性.
Aiming at the two key problems of fault detection and fault identification in model-based fault diagnosis process, a method based on neural network is proposed.Firstly, the BP neural network is used to estimate the parameters and the system model is used for fault detection. Then, The network carries on the data clustering and carries on the system breakdown identification based on the result of the clustering.Finally, the fault diagnosis system based on BP / ART2 neural network is designed and implemented.The parameter estimation method based on BP neural network can accurately estimate the diagnosis object under different states Which can provide an effective basis for fault detection. Data clustering based on ART2 neural network can not only identify the known fault types of the object, but also identify unknown faults and identify faults in systems with less prior information more effectively Through the application of the fault diagnosis case of permanent magnet DC motor, it proves that the method can have certain engineering practicability.