基于小波去噪的MEMS陀螺仪随机误差校准算法

微机电系统(micro electro mechanical system,MEMS)陀螺仪具有体积小、成本低、功耗低等特点，在微姿态测量系统中应用极其广泛。由于在制作工艺、材料等方面会引入额外的随机噪声，且瞬态电压的不稳定也会造成MEMS陀螺仪在上电阶段产生随机波动误差，严重影响微姿态测量系统的启动时间和测量精度。因此，基于多分辨分析特性的小波变换分析技术，提出了一种改进的小波去噪算法，通过对MEMS陀螺仪的数据进行3层小波分解，剔除高频分量和电压不稳定产生的突变信号，并对低频分量进行重构，最终得到校准以后的陀螺仪数据，实现陀螺仪随机误差的快速校准。实验验证结果表明，通过3层小波分解后，随机误差均值小于0.05 °/s，系统启动时间小于0.1 s，具有较好的噪声抑制和迅速启动能力。

An algorithm for random error calibration of MEMS gyroscope based on wavelet denoising

Micro electro mechanical system (MEMS) gyroscopes are widely used in micro attitude measurement systems because of their small volume, low cost, and low power consumption. However, additional random noise will be introduced in the production process and materials. In addition, the instability of the transient voltage will also cause the random fluctuation error of the MEMS gyroscope during the power-on stage, which seriously affects the startup time and measurement accuracy of the micro attitude measurement system. To solve the above problems,an improved wavelet denoising algorithm is proposed based on the wavelet transform analysis technology of multi-resolution analysis characteristics. By performing three-layer wavelet decomposition on the data of the MEMS gyroscope, the abrupt signal generated by high-frequency component and voltage instability is eliminated, and the low-frequency component is reconstructed. Finally, the gyroscope data after calibration is obtained, and the rapid calibration of the random error of the gyroscope is realized. The experimental verification results show that after three-layer wavelet decomposition, the average random error is less than 0.05 °/s, and the system startup time is less than 0.1 s, which has good noise suppression and rapid startup capabilities.