基于神经网络的MEMS陀螺仪温漂补偿方法研究

Research on temperature drift compensation method of mems gyroscope based on neural network

针对MEMS惯性传感器对温度变化非常敏感问题,提出了一种MEMS陀螺仪实时温度补偿方法,并比较了此方法与传统使用的分段线性拟合方法的补偿效果.首先,使用一个标定过的MEMS-IMU进行温度试验,并在试验过程中采集MEMS陀螺仪的输出;其次,提出了一个基于MEMS陀螺仪前一时刻温漂输出的实时补偿模型,并引入虚假最近邻算法来确定神经网络隐藏层的神经元个数;然后,采用了一种普通循环神经网络的变体,门循环单元循环神经网络,来辨识所提出模型的输入与输出之间的关系,从而得到完整的MEMS陀螺仪温漂实时补偿模型;最后,比较了所提出的方法和分段线性拟合方法的补偿效果.为了得到定量的比较结果,分别在相同试验条件下和不同升温速率条件下比较了以上两种方法补偿前后MEMS陀螺仪的艾伦方差和零偏稳定性.根据比较结果可知,在以上两种试验条件下,采用所提出的方法补偿后,MEMS陀螺仪的艾伦方差系数和零偏稳定性均显著降低,并且补偿效果均优于分段线性拟合方法.

Aiming at the problem that MEMS inertial sensors are very sensitive to temperature changes,a real-time temperature compensation method for MEMS gyroscopes is proposed,and the compensation effect of this method is compared with the traditional piecewise linear fitting method.First,an in-house-designed MEMS-IMU for temperature effect tests,and the output of the MEMS gyroscope is collected during the test;Second,a real-time compensation model based on gyroscope output is proposed at the last moment,and the false nearest neighbor algorithm is introduced to determine the number of neurons in the hidden layer of the neural network.Then,a variant of the standard recurrent neural network,which is the gated recurrent neural network,is used to identify the relationship between the input and output of the proposed model,and a complete real-time compensation model for the temperature drift of the MEMS gyroscope is obtained.Finally,We compare the proposed and the piecewise linear fitting methods5 compensation effects.Toconditions and different heating rate conditions,respectively.According to the comparison results,under the above two test conditions,the Allen variance coefficient and zero bias stability of the MEMS gyroscope are significantly reduced after the compensation by the proposed method,and the compensation effect is better than the segmented linear fitting method.Therefore,the method proposed in this paper effectively compensates for the temperature drift of the MEMS gyroscope.