在顾及动力学模型随机误差的情况下,设计了一种GPS/INS自适应滤波算法。针对BP神经网络存在的训练速度慢、易陷入局部极小值等问题,对神经网络学习算法进行了改进。利用神经网络进一步减小系统误差对导航解的影响,给出了顾及动力学模型随机误差和系统误差的GPS/INS自适应滤波算法,并利用实测数据验证了算法的有效性。
An adaptive filtering for GPS/INS integrated navigation is designed, considering the influences of stochastic errors. An improved BP neural network learning algorithm is set up, on the basis of it, a new GPS/INS adaptive filtering that combines the adjustment of stochastic errors and systematic errors is put forward. The results prove that the algorithms are valid and reliable when applied in GPS/INS integrated navigation system.