论文部分内容阅读
在采用Kalman滤波进行捷联惯导精对准时,当模型存在误差或系统噪声不能反映实际噪声时,会降低滤波精度甚至导致滤波发散。针对这个问题,提出基于Elman神经网络和Kalman滤波的捷联惯导精对准方法。首先对已知噪声统计特性的系统进行Kalman滤波,将稳定可靠的状态估值作为网络期望输出用来训练Elman网络;然后再用训练好的网络对未知噪声统计特性系统进行状态估计。利用仿真数据对该算法进行验证,结果表明,该算法能够克服Kalman滤波精对准的缺陷,提高对准精度,尤其是航向角的精度。
When Kalman filter is used for SINS alignment, when the model is inaccurate or the system noise does not reflect the actual noise, it will reduce the filtering accuracy and even cause the filtering to diverge. To solve this problem, a SINS fine alignment method based on Elman neural network and Kalman filter is proposed. Firstly, the Kalman filter is applied to the system with known noise statistics, and the stable and reliable state estimation is used as the expected output of the network to train the Elman network. Then, a well-trained network is used to estimate the state of the unknown noise statistics system. Simulation results show that this algorithm can overcome the defects of Kalman filtering and improve the alignment accuracy, especially the accuracy of heading angle.