中文摘要：

将灰色预测GM（1,1）原理引入到水下机器人传感器的故障诊断中,对传感器样本数据序列建立灰色动态预测型。通过对该模型输出信号与实际输出之间误差的分析,实时检测传感器的故障。针对Outland 1000无人水下机器人中的方向传感器,应用该方法对该传感器的三种典型故障模式进行了故障检测实验,结果表明所提故障检测方法准确可靠。

英文摘要：

The Grey prediction model theory（GM 1,1） is introduced to fault detection of underwater vehicle sensor.Firstly,the grey model is presented with sensor data of underwater vehicle;Secondly,according to the analysis of error between the output of grey dynamic prediction model and the practical sensor output signals,the sensor fault is diagnosed.The fault diagnosis method is applied in three kinds of fault patterns of compass sensor for Outland 1000 ROV.The result of the experiment proves that the proposed method is prompt and exact.