Researchfor Ball Disc Rotor Decay Gyro Signal Processing Technology Based On DSP

In the gyro operation, gyros random noise will greatly affect the performance of the gyroscope.this paper uses classical and modern filtering algorithms to reduce the random noise of the gyro of a new structure which was proposed bu Harbin Institute of Technology.Improving the performance of the gyro system is mainly related to the optimization of the gyro and signal processing.This paper researches the signal processing for the new structure gyro proposed by MEMS Center of HIT(ball disc rotor gyro).The structure of the ball disc rotor gyroscope which compared to the conventional structure is a new structure, so the previous signal processing may not be applicable.In this paper, we apply classic and modern filter to filter noise of this new structure in order to improve overall performance of the gyro.Fisr of all,this paper designs the platform of hardware which filtering algorithm runs on, in this platform, TMS320F2812 is the main arithmetic and control unit,AD7655 is the sampling chip of 16 bits which is used to collect the signal of the gyro,and on this basis, we add the DA module and DDS module to facilitate subsequent research of gyro. According to the hardware design, this paper write the drivers of each module.Final experimental measurements show that AD sampling precision can reach14 bits and the sampling rate is up to 200 k Hz.secondly,after the design of the hardware platform, this paper uses MATLAB filter design toolbox to design the IIR filter algorithm and simulate it with Simulink. Then we write this algorithms on DSP platform and optimize the it, the hardware platform of experiments show that DSP can filter out eighty percent or more of noise which is not in useful signal band in real time.Finally,classical filtering algorithms can filter out most of the noise, but it can not do anything if the noise and signal are in the band of the same frequency. Aiming at this problem this paper applies the Kalman filter. Kalman filter algorithm requires that the noise which will be processed is white noise and it has an accurate mathematical model,Therefore, this paper pretreat the collected gyro signal,we get that rounds value of the gyro signal is 96,the vale of kurtosis is 3.0149,the vale of skewness is 0.0474,and the signal meets normal distribution.Then we set up the ARMA model of gyro signal.By Kalman filter algorithm, the random signal variance is reduced from 3.0041×10-7to7.3704×10-8, effectively filtering out the random noise in the signal, greatly improving the bias stability of the gyro.