小波网络辅助卡尔曼滤波的捷联惯导传递对准

Wavelet Neural Network Aided Kalman Filter for Transfer Alignment of SINS

摘要:

初始对准精度是捷联惯导系统的主要误差来源之一;针对舰载机捷联惯导的传递对准模型准确建模困难,且测量噪声和过程噪声随舰船动态而变化,这样就会降低滤波的精度,卡尔曼滤波有一定的局限性,提出了将小波神经网络辅助卡尔曼滤波器用于惯导系统的传递对准;把能直接影响卡尔曼滤波估计误差的参数作为网络的输入,进过样本训练后,把网络的输出与经过卡尔曼滤波得到的结果相加,实现了捷联惯导的传递对准的滤波功能;这种新算法在实际应用中的非线性情况下优于传统卡尔曼滤波方法;仿真结果表明了其实用性和有效性。

Abstract:

The error of transfer alignment is one of the most important errors in strapdown inertial system(SINS). In this paper,a method based on Kalman filter aided by wavelet neural network is introduced to transfer alignment during the navigation for the limitation of Kalman filter:it is difficult to modelig the accurate transfer alignment of SINS,in which the noise of measurement and process changes by the movement of carrier-based aircraft and the accurary of Kalman filter will be affected. The filter function for Transfer Alignment of SINS is realized by adding the Kalman filter result to the sample after training which the Kalman filter estimation is as the input of the network error parameter. The proposed method provides better accuracy compared to the Kalman filter on the condition of nonlinear dynamics which is common in application. Simulation results show its effectiveness and practicality.