PSO网络优化双重卡尔曼滤波在初始对准中的应用

The Application of Double Kalman Filtering Based on PSO in SINS Initial Alignment

为了解决捷联惯导系统初始对准过程中天向方位失准角收敛过慢的问题,提出了基于带惯性权重的微粒群(PSO)优化网络逼近的双重卡尔曼滤波算法,首先,利用滤波过程中产生了新的天向方位失准角U和水平方位失准角N信息进行双重卡尔曼滤波估计。然后针对滤波结构复杂、运算量大的问题,利用双重卡尔曼滤波获取的数据作为训练样本,PSO优化神经网络的算法来实现捷联惯导的初始对准。仿真结果表明,新算法结构简单,计算量小,具有实时性和快速性,同时可以保证系统的对准精度。

In strapdown inertial navigation system （SINS） initial alignment, the estimation effect of azimuth error angle is bad. Based on this, the double-stage Kalman filtering is proposed utilizing the new information of φU and φN. To solve the complex structure and much operation, the neural network optimized by the particle swarm optimization （PSO） is used in SINS. The weigh parameters of neural network are trained by PSO algorithm with the sample data generated by double Kalman filtering. It is more attractive to use neural network to realize SINS initial alignment for simple structure, real-time working and desired precision.