基于神经网络的高超声速飞行器惯导系统精度提高方法

针对目前惯性系统误差补偿模型对静态误差和动态误差处理能力不足的问题，为适应高超声速飞行器长航时、高精度的惯性导航要求，基于神经网络提出一种加速度计拟合模型。在高超声速飞行器飞行前期有准确的卫星导航信息时，收集导航信息和加速度计脉冲信息，利用神经网络强大的非线性拟合能力，在飞行过程中进行在线训练，得到精确的惯性系统模型。仿真结果表明，在存在逐次通电误差和不考虑二次项误差系数的误差补偿模型方法位置导航偏差在数公里和数百米量级的情况下，相同时间内所提方法的位置导航偏差仅为数十米量级，有效提高了高超声速飞行器的导航精度。

Method for improving the precision of hypersonic vehicle inertial navigation system based on neural network

Aiming at the problem that the current inertial system error compensation model is insufficient in processing static and dynamic errors, in order to meet the requirements of long-endurance and high-precision inertial navigation of hypersonic aircraft, an accelerometer fitting model based on neural network is proposed. When the hypersonic vehicle has accurate satellite navigation information in the early stage of flight, the navigation information and accelerometer pulse information are collected, and the powerful nonlinear fitting ability of the neural network is used to conduct online training during the flight to obtain an accurate inertial system model. The simulation results show that when the power-on error and the error compensation model method does not consider the quadratic term error coefficient, the position navigation deviation is on the order of several kilometers and hundreds of meters, the position navigation deviation of the proposed method is only on the order of tens of meters in the same time, which effectively improve the navigation accuracy of the hypersonic vehicle.