Angular rate error compensation of MEMS based gyroscope using artificial neural network

Micro Electro Mechanical Systems based gyroscopes have been most significant inertial sensors in previous few years because of its low cost and small size. However these gyroscopes have limited usage due to its non idealities like angular rate error, temperature drift, random drift, drift periodic error. In this paper MEMS gyroscope´s angular rate error, have been studied and suitable artificial neural networks based modeling and compensation scheme is proposed. Reference angular rates and output data of MG31-300 under different input angular rates were obtained and also analyzed from literature. As the angular rate error has nonlinear behavior and randomcharacteristics, the ANN based model isdeveloped, which takes gyroscope´s output voltage with error as the inputand gives output voltage with compensation the effect of angular rate error. The model has been analyzed for number of neurons as well as mean square error. Result shows that with three number of neurons mean square error of 1.72 e-4 can be achieved which is acceptable and suitable for implementation.