Temperature Error Modeling of RLG Based on Neural Network Optimized by PSO and Regularization

Abstract

Traditional temperature error model of ring laser gyroscope (RLG) based on neural network faces great problems in the repeatability performance under different temperature conditions. To reduce the temperature error and improve the generalization ability of traditional neural network, a novel error model based on radial basis function neural network optimized by particle swarm optimization (PSO) and regularization approach was proposed. The temperature error is analyzed and preprocessed. The PSO method is used to search the optimal configuration of the network, and regularization method is used as the evaluation criterion to further optimize the coefficients of the network. The experimental results show that the proposed method can improve the precision and environmental adaptability of RLG, which had been practically applied in RLG position and orientation system.