基于BP神经网络的GPS/SINS姿态测量融合算法

针对传统全球定位系统/捷联惯性导航系统(global positioning system/strapdown inertial navigation system,GPS/SINS)组合系统信息融合技术存在的不足,设计了适用于GPS/SINS组合系统姿态测量的反向传播(back-propagation,BP)神经网络模型,并在标准BP算法基础上,采用限制初始权值和阈值?动态删减隐含层节点数和修改激活函数的改进算法,从而减小BP算法在GPS/SINS组合姿态测量过程中的训练误差,提高网络训练的相应速度及精度?最后通过算法仿真和分析结果图得出改进后的BP神经网络算法在GPS/SINS组合姿态测量系统中高效可行,且测姿性能优越?

Attitude measurement fusion algorithm in GPS/SINS based on BP neural-network

Due to the deficiency in traditional GPS/SINS system information fusion technology,this paper proposes a BP network model for GPS/SINS integrated attitude measurement system to overcome it.Based on the BP Neural-Network theory,the ways such as limiting the initial weights and thresholds,dynamically deleting the number of hidden layer node and modifying the method of activation function are made use of to extend BP algorithm''s function,and then get a high training speed and accuracy. The feasibility and effectiveness of the improved f BP Neural-Network algorithm is verified in the layout diagram.