基于无监督学习的目标轨迹预测

针对情报与侦察监视领域中目标轨迹预测问题，提出了一种基于无监督学习的预测方法。首先，根据历史信息分析目标历史活动规律；其次，构建隐马尔科夫模型，通过无监督学习自动实现预测目标在栅格网中的运动方向；最后，根据学习得到的运动方向和目标运动速度信息，计算未来短期内的目标轨迹。数值仿真验证了该方法能够有效地预测目标在未来短时刻内（通常为5 min）的运动轨迹。

Target trajectory prediction based on unsupervised learning

Aiming at the problem of target trajectory prediction in the field of intelligence and reconnaissance surveillance,a prediction method based on unsupervised learning is proposed.Firstly,a target historical activity pattern is analyzed according to historical information.Secondly,the hidden Markov model is constructed and the motion direction of the prediction target in the grid is automatically realized by unsupervised learning.Finally,The target trajectory in the short term is calculated according to the learning direction of motion and the speed of target motion.Numerical simulations show that the proposed method can effectively predict the trajectory of the target in the short time (usually 5 minutes).