基于神经网络的管制员智能冲突调配模型研究

Research on ATC's Intelligent Conflict Release Module Based on Neural Network

为解决管制冲突的智能化调配问题,通过对历史雷达轨迹数据的预处理,生成了学习样本集。建立了基于BP神经网络的管制冲突智能调配模型,并基于效果评价策略,选用样本数据对模型进行训练。搭建了多Agent空中交通仿真平台,在管制员Agent中嵌入智能调配模型,通过冲突解脱场景仿真验证了智能冲突调配模型。实验结果表明:模型可以较好地根据当前冲突情况从历史经验中寻找解脱的方案,实现冲突解脱。

To solve the problem of intelligent conflict control deployment,the paper generates the learning sample set by data preprocessing of history radar track. We establish the intelligent control conflict release model based on BP neural network and select sample data to train model according to the effect evaluation strategy. We set up the multiple Agent air traffic simulation platform,the intelligent conflict control deployment model is embedded in the ATC Agent and is verified by conflict release scene simulation. The experimental results show that the model can find the conflict release well from the historical experience,according to the current conflict situation.