摘要
小波神经网络具有收敛速度快、结构简单、计算量少等优点。该文首先论证了小波神经网络的理论基础,然后给出了小波神经网络的参数估值方法及隐层小波元个数的确定依据,并将其用于惯导初始对准中。仿真结果表明,该方法能有效地实现初始对准的状态估计,既得到了与卡尔曼滤波器相当的精度,又减少了卡尔曼滤波的过渡时间,提高了系统的实时性及收敛性。
The wavelet neural network(WNN)has the advantage of fast convergence,simple structure and small computation etc.Firstly,the mathematical foundation of WNN is presented,secondly the estimation of parameter and the quantity of wavelet element are analyzed,then the method have been used in initial alignment.Simulation results indicate that the method can efficiently realize the estimation of initial state.Not only can it obtain the precision similar to that of kalman filter,but also can reduce the time of kalman filter process and improve the system's real time and convergence.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第30期215-217,229,共4页
Computer Engineering and Applications
