一种GRNN神经网络的高超声速飞行器轨迹预测方法

HYPERSONIC VEHICLE TRACK PREDICTION BASED ON GRNN

在针对高超声速飞行器的拦截作战中,对其轨迹的准确预测是成功拦截的重要基础。基于对高超声速飞行器轨迹特性的分析,以广义回归神经网络GRNN(Generalized Regression Neural Network)理论为依据,对高超声速飞行器的轨迹预测进行研究。通过对轨迹误差产生的原因进行分析,针对高超声速飞行器提出了基于误差修正的GRNN轨迹预测算法。通过仿真验证了算法的有效性,研究结果表明,随着观测数据的增加,算法的误差范围逐步减小,误差修正的效果越来越好。

In the combat of hypersonic vehicle defense, the precision of track prediction is an important basis to successiul deiense. Based on the analysed characteristic of hypersonic vehicle track and taking the GRNN theory as basis, in this paper we studied the track prediction of hypersonic vehicle. By analysing the cause of track error generation, we proposed the error correction-based GRNN track prediction algo- rithm for hypersonic vehicle. The algorithm was validated through simulation, research result showed that the error range of the algorithm de- creased gradually and the effect of error correction got better and better along with the increase of observation data.