基于MAS和BP网络的近距离目标预测法

Near Distance Target Prediction Method based on MAS and BP Network

以轨迹拟合法为基础,将目标预测问题转化为对预测模型参数优化搜索的问题。结合MAS、BP网络和位置预测原理,提出将近距离目标视为刚体。其状态按自由度划分为6个,其中3个自由度表示空间位置,另外3个自由度表示角的转动向量,6个自由度看作6个子网,分别进行子网预测,再进行组网的近距离目标预测法。最后,以无人机为例对算法进行仿真计算,验证了其正确性和可行性。

The target prediction problems were changed into searching optimized parameters problems of prediction mod- el based on the trajectory fitting method. The near distance target prediction method was given by dividing the rigid near dis- tance target state into six according to the freedom degrees, three freedom degrees of them were space position, the other three freedom degrees were angle rotation vector, 6 freedom degrees were look as six subnet and were separately done sub- net prediction, and then networking were proceeded based on the MAS, BP network and position prediction theory. Finally, correctness and feasibility of the algorithm were tested by the simulation and calculate of UAV.