基于BP神经网络的光纤陀螺仪温度建模研究

Temperature modeling study for FOG based on back-propagation neural network

目前光纤陀螺应用广泛,但是其性能容易受到环境温度影响,从而影响到惯性导航系统的性能。光纤陀螺的温度特性具有非常复杂的非线性特点,而BP神经网络具有良好的逼近复杂非线性函数能力。使用BP神经网络建立光纤陀螺温度特性的黑箱模型,不对零漂和标度因子进行补偿,而直接对陀螺输出进行校正。经实际数据检验,该建模补偿方法比未经补偿和经过传统工程补偿方法的精度提高了两个数量级。与传统的线性模型相比较,本文基于BP神经网络建立的光纤陀螺温度模型具有补偿方法简单,精度高,通用性好等优点。

Fiber Optic Gyroscopes （FOG） is widely used, but is easily affected by the temperature around it, which affects the performance of the inertial navigation system. Its temperature model has a very complicated nonlinear characteristic. Back Propagation （BP） neural network has the advantage of approximating the nonlinear function. The black-box temperature model of FOG can be modeled by using BP network and directly correct the error of the output of FOG without compensating gyro drift and scale factor as the traditional method does. Tested by the experimental data, the accuracy of the black-box model based on BP neural network is improved by two orders of magnitude, compared to the uncorrected output and the traditional method. Compared to the traditional linear model, it has more advantages, such as easy to use, high precision, good applicability, etc.