灰色动态预测在AUV传感器故障诊断中的应用

Application of Sensor Fault Diagnosis Based on Grey Dynamic Prediction for AUV

针对自主水下机器人(AUV)传感器故障诊断中样本数据少、随机性强、实时性要求高的特点,将灰色动态预测模型的建模原理引用到AUV传感器的故障诊断中。在对传感器进行数据滤波、小样本灰色建模与灰色动态预测的基础上,可以实现AUV传感器的实时故障诊断。文章详细阐述了基于灰色动态预测的传感器故障诊断的具体实现方法和步骤,对AUV传感器中典型的四种故障模式进行了仿真研究。结果表明该方法能快速、准确地诊断出传感器故障,并且在传感器发生故障后的一段时间内能够实现信号恢复。

According to the properties of short of information, strong randomicity and real-time requirement in sensor fault diagnosis for Autonomous Underwater Vehicle （AUV）, the modelling principle of grey dynamic prediction method is introduced. The real-time fault diagnosis of AUV sensor can be fulfilled based on data filtering, grey modeling of small-scale sampling and grey dynamic prediction. The detailed practicing method and steps of sensor fault diagnosis based on grey dynamic prediction are proposed and the simulation research is carried out for four typical fault modes of AUV sensor. The result shows that the method can diagnose the sensor faults fast and accurately, and can recover the signal after faults happening for a period of time.