基于卡尔曼滤波算法展开的飞行目标轨迹预测

Flight target track prediction based on Kalman filter algorithm unfolding

目标轨迹预测是保证目标航行安全、规划飞行航迹和搜寻空中目标等任务的关键技术,在军事和交通管制等方面具有重要意义。针对传统飞行目标轨迹预测方法模型较为简化且预测精度较低的问题,提出了基于卡尔曼滤波算法展开的深度神经网络模型,用于飞行目标的轨迹预测任务。该模型通过长短时记忆(long short-term memory, LSTM)网络从目标的航迹数据中学习目标的运动状态,再利用卡尔曼滤波算法对LSTM预测的目标状态估计值进行动态修正,其有效结合了卡尔曼滤波算法和深度神经网络各自的优势。在仿真数据和真实数据上的实验验证了所提模型较其他网络模型对飞行目标轨迹预测的准确性和有效性优势。

Target track prediction is a key technology to ensure the navigation safety of the target,to plan the flight path and to search the air target.It is of great significance in military and traffic control.Aiming at the problem that the traditional air target track prediction method model is relatively simplified and the prediction accuracy is low,a deep neural network based on the Kalman filter algorithm for the air target track prediction task is proposed.The model uses long short-term memory(LSTM)network to learn the target motion model from the track data of air targets,and then uses the Kalman filter algorithm to dynamically modify the estimated target state generated by LSTM,which effectively combines the advantages of the Kalman filter algorithm and deep neural network.Experiments on simulation data and real data verify the accuracy and effectiveness of the proposed model compared with other networks models for air target track prediction.