基于混合智能系统的设备故障诊断研究

Research on equipment fault diagnosis based on hybrid intelligent systems

为研究和改进人工智能技术在设备故障诊断中的缺点和不足,提高故障诊断的准确率,构建了一种混合智能诊断系统。首先利用小波包分析技术对设备故障进行特征提取和分析;接着对数据进行离散化处理,应用粗糙集对获得的故障特征向量进行约简,删除冗余信息;然后利用免疫遗传算法的全局优化能力去训练BP神经网络的权值,建立免疫遗传-BP神经网络模型;最后把经粗糙集约简后的故障特征向量输入该模型,完成故障识别和智能诊断。通过旋转机械的转子系统的仿真实验,表明基于小波包-混合智能的故障诊断取得了良好的诊断效果。

To improve shortcomings and the insufficiency in the research of artificial intelligence technology in equipment fault diagnosis,and to enhance the accuracy in fault diagnosis,a hybrid intelligent diagnosis system is constructed.First,wavelet packet analysis technique is used for feature extraction and analysis of equipment failure.Second,the data is dispersing processed,application of rough set to obtain fault characteristic vectors reduction,delete redundant information.Third,global optimization ability of immune genetic algorithm is then used to train the weights of BP neural network,immune genetic-BP neural network model is set up.Finally,after the input of the fault feature vectors by rough set reduction,the fault recognition and intelligent diagnostics is completed.Through the simulation experiment of rotating machinery rotor system,suggests hybrid intelligent fault diagnosis based on wavelet packet has obtained the good diagnosis effect.