基于神经网络预测器的单传感器故障检测方法

目的 研究单传感器的故障检测方法，方法 提出了神经网络预测器用于单传感器故障检测的实现方法，并给出了神经网络预测器的在线和离线学习算法。结果 本方法只需要一个传感器的输出信号就可检测系统中的传感器是否发生故障，如果对系统中多个传感器建立神经网络模型，可以检测控制系统中多种类型的以及多个同时发生的传感器故障。结论 用汽车发动机仿真实验可以表明本方法是行之有效的。

Detecting Single Sensor's Fault Based on Time Series Predictor Using Neural Network

Aim To study the detection of the single sensor's fault by using its output signal. Methods The principle of neural network based on time series predictor was presented. The on line and off line training algorithms of the predictor were summed up. Results The method has many advantages over other methods such as the ability to detect sensor fault by using the single sensor's output signal, the ability to detect multiple sensor faults, the ability to detect two sensor faults occurring simultaneously and the ability to detect many kinds of sensor faults. Conclusion The simulation results through an automotive engine model show that the method can detect the sensor fault successfully.