基于小波模糊神经网络的陀螺仪故障诊断技术

Application of Wavelet Fuzzy Neural Networks for Fault Diagnosis

为了提高陀螺仪故障检测灵敏度 ,分析了陀螺漂移特性 ,指出随机漂移是影响陀螺仪精度的主要漂移误差 ,也是影响性能可靠性的主要因素 .针对此 ,提出了一种小波模糊神经网络故障诊断模型 .此模型运用串联方式将小波分析、模糊逻辑和神经网络融合在一起 ,充分发挥它们各自的优点 ,并对陀螺仪实测数据仿真 .仿真结果表明 ,采用小波分析提取陀螺的常值漂移 ,简单有效 ;运用神经网络对陀螺的误差和故障建模 ,使故障诊断具有自适应、自学习的能力 ,并且增强了故障诊断的容错能力 ;将模糊逻辑用于判决中 。

Gyro's fault diagnosis plays a critical role in inertia navigation system for higher reliability and precision. Thus, in order to improve fault detection sensibility, gyro's drift characteristics was analyzed and it is pointed out that the random drift is not only the main drift error which influences gyro's precision, but also the primary factor affecting gyro's reliability. Based on the analysis of the existing methods, this paper developed a diagnostic model which seriesly integrates the three promising technologies: wavelet transform, artificial neural networks (ANNs) and fuzzy logic. Simulation was performed with real gyro's readings to evaluate the proposed diagnostic model. The results show: (1) using wavelet transform to extract the trend item of gyro drift is very simple and effective; (2) employing ANNs to build model of gyro's error and fault makes fault dignosis possess capabilities of self-study and self-adaption, and enhances error-tolerance of failure diagnosis; (3) fuzzy logic judgement lets diagnostic results more reliable. The proposed diagnostic model increases the accuracy of the fault diagnosis and can be generalized to other devices' fault diagnosis and identification in the navigation systems.