文献类型：期刊文献

中文题名：基于融合神经网络的传感器故障诊断专家系统

英文题名：Robust Fault Detection Method of Aircraft Engine Sensors

作者：魏红娟[1]

第一作者：魏红娟

机构：[1]新乡学院计算机与信息工程学院

第一机构：新乡学院计算机与信息工程学院

年份：2013

卷号：21

期号：2

起止页码：358-361

中文期刊名：计算机测量与控制

外文期刊名：Computer Measurement & Control

收录：CSTPCD;;北大核心:【北大核心2011】；

语种：中文

中文关键词：神经网络;传感器;故障诊断

外文关键词：Neural network; Sensor; Fault diagnosis

摘要：针对传感器抗振动差、易受干扰的故障特点,常规采用的基于模型的传感器故障检测方法的准确性易受到建模误差与外界扰动的影响,造成漏报或误报;提出了一种基于融合神经网络的传感器故障诊断专家系统;该方法采用常规电流残差信号和振动幅度信号作为传感器故障诊断的输入信号,应用融合BP神经网络进行传感器的故障方法,分别用局部故障诊断模型,对局部的传感器故障信息进行全局决策的融合,从而提高专家故障诊断系统的准确率;研究结果表明,该传感器故障专家系统具有诊断准确率高、诊断速度快等优点,抑制干扰对故障检测的影响,诊断准确性超过75%。
According to the sensor shock resistant difference, susceptible to interference fault characteristics, based on the conventional model of sensor fault detection method accuracy susceptible to modeling error and the influence of the external disturbances, causing the o mission or false positives. Puts forward a neural network based on the fusion of the sensor fault diagnosis expert system. This method USES conventional current residual signal and vibration amplitude signal as sensor fault diagnosis of input signal, the application of fusion BP neural network for sensor fault method, respectively for local fault diagnosis model for local sensor fault information fusion of global decision, so as to improve the accuracy of fault diagnosis expert system. Research results show that the sensor fault diagnosis expert system has high accuracy, and the diagnosis speed etc, and suppress the interference to the influence of fault detection and diagnosis accuracy of more than 75 %.