基于统计参数分析和RBF网络的动调陀螺故障诊断方法

A Fault Diagnosis Method for DTGs Based on Statistical Parameter Analysis and RBF Neural Network

针对动调陀螺故障振动信号的特点,提出了一种基于振动统计参数分析和神经网络的动调陀螺故障诊断方法。该方法通过计算原始振动信号的一组统计参数作为表征故障的特征信息,以此作为RBF神经网络的输入参数来学习并识别陀螺故障。实验结果表明,采用对统计参数的计算能够简单、有效地提取陀螺故障特征信息;运用神经网络进行故障诊断建模,使诊断具有自适应、自学习的能力,诊断结果更加可靠。

Fault diagnosis of gyroscopes plays a critical role in inertial navigation system for higher reliability and precision. In this paper, the statistical parameter analysis, a kind of time domain analysis approach for vibration signal, is introduced and a fault diagnosis method based on the statistical parameter analysis and RBF neural network is proposed for dynamically tuned gyroscopes （DTG）. This method first employs the statistical parameter analysis to compute a set of statistical parameters of vibration signal, with which the RBF neural network is then constructed to train and identify the working state of the DTG. The experimental results verify that the proposed diagnostic model can simply and effectively extract the state feature of DTG and is reliable and practical.