基于自适应神经模糊推理系统的陀螺建模方法

Modeling of gyroscope based on adaptive neuro-fuzzy inference system

系统地介绍了近年来陀螺信号处理中出现的各种主要的滤波方法,并分析了各自的优缺点。针对陀螺输出信号自身的特点,提出了一种基于自适应神经模糊推理系统的陀螺滤波方法,建立了陀螺误差模型。提出的滤波方法同时具有神经网络与模糊逻辑推理的优点,从而有效地减小了建模误差。分别探讨了陀螺静态时的零偏信号估计模型与敏感动态信号时的滤波方案。由仿真结果可以看出,该方法可以有效地估计陀螺的零偏,并能抑制陀螺输出信号中的噪声。

The advantages and disadvantages of the algorithms developed recently in the field of filtering of gyroscope were discussed. A new filtering method for gyroscope was presented based on the Adaptive Neuro-Fuzzy Inference System, and the model of gyroscope was established. The new filtering has both the advantages of neural network and fuzzy logical system, in which the modeling error is reduced. The filtering schemes of zero bias in the situation of no input and the output of gyroscope when input exists were discussed respectively. The modeling results show that the zero bias can be estimated properly, and the noise is eliminated effectively.