【摘要】 传感器故障诊断在化工生产中有着重要地位。该文以小波变换与神经网络方法为基础,提出了一种传感器故障诊断的方法。该方法能够有效区分传感器故障造成的信号变化与过程本身正常波动造成的信号变化,同时在训练神经网络时只需要系统正常状态下的样本,克服了传感器故障样本稀少的困难。此外,该方法可以在传感器发生故障后估计出正常的模拟信号。实验证明,该方法能够有效完成故障诊断,并可以判断出传感器的故障类型。更多还原

【Abstract】 Sensor fault diagnosis plays an important role in chemical engineering. This paper proposes a method for sensor fault diagnosis based on wavelet transform and neural network, which can distinguish signal change caused by sensor fault from normal process dynamics. Furthermore, this method needs only the samples of the system under normal situation while training the neural network and overcomes the difficulties caused by the lack of sensor fault samples. This method can also calculate the normal simulation signal when there is a sensor fault. After a series of experiments, this method is proved to be applicable in fault diagnosis and can identify the fault types.更多还原