MEMS Gyroscope Temperature Compensation Based on SSA-RBF Neural Network

Abstract

The output of the Micro Electro-mechanical System (MEMS) gyroscope is susceptible affected by temperature drift, which reduces the measurement accuracy of the gyroscope. In this paper, a gyroscope temperature compensation method based on sparrow search algorithm (SSA) and radial basis function (RBF) neural network is proposed to reduce the temperature drift error of gyroscope. Firstly, we utilize the RBF neural network to establish the model of temperature error on the original output of gyroscope; then SSA is employed to find the optimal parameters of the RBF neural network in order to improve its search speed and generalization performance; finally, the optimized RBF neural network is applied to the temperature compensation of the gyroscope. The numerical simulation and comparison results under different temperatures demonstrate that, compared with polynomial and RBF neural network, the SSA-RBF neural network compensation method has superior compensation accuracy and faster convergence speed, which significantly reduces the maximum error, mean value and the standard deviation of gyroscope. Thus, the proposed SSA-RBF method can obtain more accurate fitting performance, effectively compensate the temperature error of MEMS gyroscope, and improve the MEMS gyroscope measurement accuracy.