基于RBF网络的高炉热流分析传感器故障诊断

在炼铁高炉热流强度分析系统中要用到温度、流量等传感器,为确保热流分析系统中传感器数据的可靠性及系统的连续、稳定运行,诊断系统用径向基函数(RBF)神经网络对传感器进行故障判断。系统由上位机、温度及流量采集装置、传感器等组成,采用RBF神经网络为每一个传感器建立预测模型,网络的输入为传感器采集信号最近的n个值,输出为该传感器在n+1时刻的预测输出值。网络通过在线学习实现对传感器的在线故障监测,经仿真分析表明:用RBF神经网络构建预测模型可满足实时性的诊断要求,提高了诊断系统的诊断精度。

Fault diagnosis for heat quality analysis sensor of puddling furnace based on RBF neural network

Temperature or flux sensors may be used in the the heat quality intensity analysis system for puddling furnace,RBF neural network is used to judge sensor fault for insureing the credibility of sensor data and the stability running of the heat quality intensity analysis system.The system is composed with computer,temperature and flux data acquire installation and sensor.Forecast model for every sensor using RBF neural network is found.The number of lately input signal for the net is n,the output of the net is the forecast value at n+1 occasion.The fault diagnosis of sensor is carried out by the on-line learning method.Emulation analysis shows forecast model for every sensor using RBF neural network can fulfil fault diagnosis and improve the diagnosis precision.