摘要
微机械陀螺仪在不同温度下工作体现出的性能有很大的不同,温度特征的存在,阻碍着陀螺的发展和应用,是误差的主要来源.为此本文提出了一种基于样本熵(SE)、变分模态分解(VMD)、BP神经网络、粒子群算法(PSO)以及时间频率峰值滤波(TFPF)的温度补偿方法.温度实验结果证明了该方法的优越性,经过算法处理后,补偿信号的零偏稳定性为1.089°/h,角速度随机游走为0.01815°/h√Hz,相比于原始信号的25.07°/h和角速度随机游走信号的0.94867°/h√Hz,该方法分别优化了95.66%和98.09%.该方法是平行处理方法,将陀螺输出中不同程度噪声和漂移进行分别处理,有效降噪的同时为神经网络提供了良好的数据学习.
The performance of micromechanical gyroscopes working at different temperatures is different.The existence of temperature characteristics hinders the development and application of gyroscopes and is the main source of errors.Therefore,this paper proposes a temperature compensation method based on sample entropy(SE),variational modal decomposition(VMD),BP neural network,particle swarm optimization(PSO)and time-frequency peak filter(TFPF).The temperature experiment results prove the superiority of the method.After processing by the algorithm,the bias stability of the compensation signal is 1.089°/h,and the angular velocity random walk is 0.01815°/h√Hz,which is compared with the original signal of 25.07°/h and 0.94867°/h√Hz are optimized by 95.66%and 98.09%respectively.This method is a parallel processing method.Different degrees of noise and drift in the output of the gyro is processed separately,which can effectively reduce noise and provide good learning data for the neural network.
作者
郭文静
曹慧亮
GUO Wenjing;CAO Huiliang(Department of Electronic Engineering,Taiyuan Institute of Technology, Taiyuan 030008, China;School of Instrument and Electronics, North University of China, Taiyuan 030051, China)
出处
《测试技术学报》
2021年第6期495-502,共8页
Journal of Test and Measurement Technology
基金
国家自然科学基金资助项目(51705477)。
关键词
MEMS陀螺仪
温度补偿
变分模态分解
BP神经网络
粒子群优化
MEMS gyroscope
temperature compensation
variational mode decomposition
BP neural network
particle swarm optimization
