Compensation of Temperature Drift of MEMS Gyroscope Using BP Neural Network

In order to solve the temperature change on the impact of MEMS gyroscope output, through self-development MEMS IMU temperature test, we used BP neural network to predict the temperature drift of MEMS gyroscope and compensation it. After using BP neural network to compensate three gyroscopes, we compared the results with the traditional least-squares method that the output error variance respectively reduced from 0.4242, 0.3506 and 0.4335 to 0.0758, 0.1024 and 0.1122. The results indicate the correctness and validity of the method, so as to offer a new method to improve the performance of MEMS gyroscope.