采用神经网络的传感器故障诊断方法研究

Study on Instrument Failure Diagnosis Using Arstificial Neural Networks

使用专用观测器组（ＤＯＳ）的方法能解决传感器故障诊断问题。然而，对于对象为非线性的、对象模型的结构和参数均未知的系统，ＤＯＳ很难构造出来，本文使用多层神经网络实现了ＤＯＳ以及决策逻辑，从而实现了对不确定的非线性系统传感器故障诊断。

The problem of instrument failure diagnosis(IFD)can be solved using dedicated observer scheme( DOS).But it is difficult to construct DOS where the process is nonlinear and the structure and the parameter of its model are all unknown.In this paper,DOS and the decision logic are achieved using the multilayered neural networks,accomplishing IFD for unknown nonlinear system.