A temperature compensation model of MEMS gyro based on BP neural network is proposed in this paper. According to recent researches, the temperature error accounts for about 80% of the total MEMS gyro error. So, it is an effective way to improve the precision of MEMS gyro by compensating the temperature error. However, the temperature error is a nonlinear characteristic, therefore, it can only be estimated by experiment. Traditional modelling methods always use piecewise linear function to fit the temperature error model, they can get accurate fitting results at segmentation points. But for other temperature points, the compensation cannot produce effective results. Based on the BP neural network, which has powerful ability on fitting nonlinear functions, a better model for temperature error of MEMS gyro is built in this paper by analysing amounts of data from repeated experiments. The results show that the method can fit the curve of temperature error well, and has better accuracy comparing to traditional methods. Meanwhile, the method is extensible and valuable in engineering practice fields.
展开▼
