Predictive filters with model error estimations using neural networks 用神經網絡估計模型誤差的預測濾波算法
As other predictive filters , state space is recursively got from measure space with system model by using the particle filter 這種濾波和其他預測性濾波一樣,可以通過模型方程由測量空間遞推得到狀態空間。
For kalman filter will tend to instability if dynamical model is incorrect , two algorithms is developed named predictive filter , which are based on kinematics model and dynamics model respectively 針對模型的不準確帶來的濾波性能下降甚至發散,提出了分別基于運動學模型和動力學模型的預測濾波算法。
According to the non - real time phenomena of high dynamic flight gps application , kalman predictive filter is designed to process the positioning data so as to reach the real - time demands 本文分析了高動態情況下gps誤差源對定位精度的影響,并且對全彈道gdop進行仿真,以此說明接收機最佳工作時段的選擇。
Simulation result indicated that predictive filter of the dynamics model may be followed the attitude movement , if disturbance moment is time variable and large relative to gradient moment of gravity , and predictive filter based kinematics is suitable if there is no information about dynamics 在有較大時變干擾力矩情況下基于動力學模型的預測濾波算法能夠很好地跟蹤姿態變化,而在對系統模型一無所知,系統動態不高的條件下利用基于運動學的預測濾波算法可以得到較好的姿態估計。