BP神经网络在惯导系统传递对准中的应用

Application of BP Neural Network to Transfer Alignment of INS

摘要:

针对系统阶次较高时卡尔曼滤波实时性较差的特点,将多层BP神经网络替代卡尔曼滤波应用于舰载机惯导系统的传递对准。利用卡尔曼滤波的输入、输出作为BP神经网络滤波的样本对值进行训练,得到了神经网络的输出值,实现了惯导传递对准中的滤波功能。仿真结果表明,将BP神经网络用于传递对准,既获得了与卡尔曼滤波相当的精度,又有效地降低了系统的解算时间,提高了系统的实时性。

Abstract:

Considering the character of Kalman filter’s bad real time when system step is high, multi-layer BP neural network is applied to transfer alignment of carrier aircraft’s INS. Training the sample of the input and output of Kalman filter, the output of neural network was obtained, and the function of estimation of transfer alignment in the INS was realized. Simulation results show that with the application of BP neural network in transfer alignment, not only get the precision similar to Kalman filter, but only reduce calculating time of system efficiently and raise the quality of real time.