基于跟踪微分器的光纤陀螺温度误差建模方法

Modeling method of gyroscope temperature error based on tracking differentiator

光纤陀螺测量精度不仅受环境温度的影响,也受温度变化率影响。针对在实际环境中温度复杂多变,温度变化率不易准确测得,而采用传统微分器得到温度变化率会放大噪声从而造成数据严重失真这一系列问题,提出了利用跟踪微分器对输入信号进行跟踪并输出微分信号的方法。首先利用跟踪微分器对采集到的温度数据进行处理,得到温度和温度变化率数据,然后利用径向基函数(RBF)神经网络强大的非线性建模能力,将温度和温度变化率作为神经网络的两个输入样本、陀螺漂移作为期望输出进行温度建模,最后对光纤陀螺温度漂移进行补偿校正。为验证算法有效性,设计了陀螺仪温度漂移实验,对实测数据进行仿真研究,结果证明跟踪微分器可以有效且快速跟踪实测温度,并输出较为真实的温度变化率曲线;利用神经网络对温度漂移进行补偿后,陀螺漂移的均值减小了两个数量级,标准差也有所减小,补偿后标准差达到6. 305E-06。

Fiber optic gyroscope is not only affected by the ambient temperature, but also the temperature change rate. In the actual environment, the temperature is complicated and the temperature change rate can not be accurately measured. If using the traditional differential for temperature change rate will amplify noise, resulting in serious distortion of data. To solve the above problems, a tracking differentiator was used to track original signal and output differential signal. Firstly, the tracking differentiator was used to process the collected temperature data to obtain the temperature and temperature change rate data. And then by using the powerful ability of nonlinear modeling of Radius Basis Function （RBF） neural network, the temperature and temperature change rate were used as two inputs of the neural network, the gyroscope drift as the expected output, the temperature was modeled. Finally, the temperature drift of gyroscope was compensated and corrected. To validate the algorithm, the experiments about temperature drift of gyroscope were designed and the measured data was simulated. The results prove that the tracking differentiator can effectively and quickly track the measured temperature and output the real temperature change rate curve. Compensated for temperature drift with the neural network, the mean of gyroscope drift is reduced by two orders of magnitude, and the standard deviation is reduced to 6. 305E - 06.