Modeling and compensation of MEMS gyroscope output data based on support vector machine

In this paper, the output data of MEMS gyroscope MG31-300 under different input angular velocity is collected and analyzed. Aiming at the nonlinear and random characteristics of MG31-300 output error, the support vector machine model of MG31-300 output data is established, and the model is trained by experimental test data to determine the model parameters. This model fits the MG31-300 output error well, and the model fitting error was 0.138°/s (1σ). A number of the angular velocity test points outside the training samples are selected to test and verify the model. The results indicate that, within MG31-300 measuring range, the angular velocity measuring accuracy is increased by an order of magnitude with this model. Therefore the support vector machine model has high precision and good generalization ability.