摘要
针对光纤陀螺零偏漂移随温度呈复杂的非线性变化,建立了BP-AdaBoost(Back Propagation neural network,Adaptive Boosting)模型对零偏进行补偿,改善了光纤陀螺的零偏稳定性能.同时,研究了模型参数对预测精度的影响,给出了BP神经网络隐含层神经元个数的选择以及AdaBoost模型迭代次数的确定方法.运用AdaBoost算法提升单个BP神经网络的预测能力,提高了集成模型整体的预测精度.对采集的光纤陀螺输出实测数据进行了事后仿真,结果表明,BP-AdaBoost模型相比传统的线性回归模型、混合线性回归模型、单个BP神经网络模型的补偿效果更显著,验证了该模型的有效性,具有重大的工程应用参考价值.
Aimed at the complex nonlinearity in bias temperature error model of fiber optic gyroscope （FOG）, based on back propagation （BP） neural network and adaptive boosting（AdaBoost） learning algo rithm, a new BPAdaBoost temperature compensation method was proposed to effectively enhance the FOG bi as stability. The effects of model parameters on the prediction precision were also investigated. A program for determining the number of hidden layer neurons in BP neural network and the number of iterations in AdaBoost model was given. The prediction error by this BPAdaBoost algorithm is smaller than that by single BP neural network. By large amount of experiments and calculations from FOG, the compensation results show that, the proposed approach has better performance compared with those traditional linear regression model, mixed line ar regression model, and single BP neural network. Through the analysis and simulation, this approach im proved is validated and has a great value of engineering reference.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2014年第2期235-239,共5页
Journal of Beijing University of Aeronautics and Astronautics
基金
国家安全重大基础研究资助项目(613186)
中央高校基本科研业务费专项资金资助项目(YWF-10-01-B30)
关键词
光纤陀螺
温度补偿
ADABOOST算法
BP神经网络
fiber optic gyroscope
temperature compensation
adaptive boosting （AdaBoost） algorithm
back propagation （BP） neural networ]~
