摘要
讨论基于神经网络预测器的传感器故障诊断问题。介绍了传感器故障诊断技术的发展,提出了一种基于神经网络在线学习的传感器故障实时诊断的模型。通过比较三种前馈神经网络的预测残差确定网络类型。介绍了网络的学习规则说明了在线学习的能力。最后,通过电厂高加热器的几个温度传感器的实际数据为例说明了此方法的实效性。
Sensor fault diagnosis based on neural network predictor is discussed. The development of the technology for sensor fault diagnosis is introduced . A kind of module for sensor fault diagnosis with the neural network with the ability of on-line learning is proposed. By the contrast of the residuals of three forward neural networks, one of them is the best. The learning rule of the selected one is introduced, and the ability of on-line learning is presented. Finally, the case of some temperature sensors in the high-level heater of the power plant shows the effectiveness of this approach.
出处
《传感技术学报》
CAS
CSCD
北大核心
2005年第2期235-237,共3页
Chinese Journal of Sensors and Actuators
