针对三容水箱是典型的非线性多变量耦合系统,提出将提升小波和概率神经网络相集合(LWPNN)方法用于三容水箱液位故障诊断.首先,建立三容水箱数学模型以获取实验数据,之后利用提升小渡对数据快速降噪,最后通过概率神经网络进行故障分类.实验结果表明,LWPNN集合方法较传统PNN故障诊断方法有较高的故障诊断率,是一种有效的故障分类方法.%An ensemble fault diagnosis approach based on lifting wavelet and probabilistic neural network (LWPNN) for three -tank system, a typical nonlinear multivariable coupled system, is presented in this paper. First, mathematical model of the three - tank system is established to get experimental data. Then the data are pre - processed to remove noise by lifting scheme wavelets. Finally, PNN is used in fault classification. Experimental results show that LWPNN is more accurate than traditional PNN fault diagnosis method, and it is an effective method of fault classification.
展开▼