Research On MEMS Gyroscopes Signal Processing Technology Based On DSP

MEMS gyroscopes, with the characteristics of small volume, light weight, low cost and high reliability, developes very quickly and gets more and more application in the low cost navigation areas. However, owing to the manufacturing process and design level, it causes several problems, such as big random noise, poor zero stability, and vulnerable to temperature affects, which influences the accuracy of navigation system. Usually, there are two methods to reduce the influence of these factors: One is to optimize the manufacturing process and MEMS gyroscope design structure, the other is to start from signal acquisition, signal denoising and temperature compensation. To achieve the purpose discussed above, the dissertation includes several works as follows:Contrive a system scheme of MEMS gyroscopes signal processing, which introduces the minimum system of TMS320F28335 in detail, as well as the hardware interface design. Hardware system includes extend RAM interface, A/D interface, extend FLASH interface, at the same time the whole design scheme is proposed.In order to surmount the real-time shortcoming of hypercycle program structure, the system uses the micro embedded operation system named DSP/BIOS, which configs and optimizes the system task, such as system initialization, signal acquisition and processing, data storage and transmition, which improves the real time of system.At present, lowcost MEMS gyroscopes has the characteristics of big random noise, poor zero stability, and vulnerable to temperature affects. A real-time filtering algorithm that combines moving average filter and median filter is proposed, then use the theory of wavelet denoising to eliminate high frequency noise in gyro signal. Meanwhile, the method of the MEMS gyro signal calibration is studied, and the temperature signal is integrated into the real-time acquisition gyro signal, a compensation algorithm that compensates gyro drift error from temperature changing is achieved. Experiment shows that the SNR increases from 121.5461dB to 150.0964dB, and temperature compensation can effectively restrain the trend term caused by temperature. The algorithm proves to be practical and effective, which has some practical value.