基于LSTM的机场飞行区活动目标潜在冲突预测

Potential conflict prediction of mobile targets in airfield areas based on LSTM

针对机场飞行区冲突不断的问题,提出一种基于长短期记忆(LSTM)网络预测机场飞行区活动目标潜在冲突的方法。根据复杂网络理论,以航空器和车辆2类活动目标为研究对象,建立飞行区活动目标网络,设置网络动态演化模型,输入运行数据计算多个网络特征指标,对指标时间序列进行主成分分析,拟合成潜在冲突指数;利用Keras框架搭建LSTM网络模型,将指标时间序列输入LSTM网络进行训练和预测,并与其他预测方法对比;用西安咸阳机场实际运行数据进行实验,将预测值与真实值进行对比,各项指标预测均方误差分别为1.608%、13.126%、0.072%、0.004%、0.014%。结果表明:通过建立飞行区活动目标网络模型,可以用网络特征指标从不同角度刻画潜在冲突;LSTM网络可以有效预测飞行区活动目标网络的潜在冲突,提醒相关人员预防冲突发生,降低冲突概率。

In view of the problem of frequent conflicts in airfield areas,a method to predict the potential conflicts of mobile targets in airfield areas based on long short-term memory(LSTM)network was proposed.According to the complex network theory,aircraft and vehicles were taken as the research objects,and the network of mobile targets in the airfield area was established.The dynamic evolution model of the network was set,and the operation data was input to calculate multiple characteristic indicators of the network.In addition,the principal component analysis of the indicator time series was carried out to synthesize the potential conflict indicator.A LSTM network model was built by using the Keras framework,and the indicator time series were input into LSTM network for training and prediction and compared with other prediction methods.The actual operation data of Xi’an Xianyang Airport were used for experiments.The predicted values were compared with the real values.The mean square errors of the predicted results of each indicator were 1.608%,13.126%,0.072%,0.004%,and 0.014%,respectively.The results show that the potential conflicts can be described from different perspectives by using characteristic indicators of the network after the network model of mobile targets in the airfield area is built.LSTM network can effectively predict the potential conflicts in the network of mobile targets in the airfield area,remind relevant personnel to prevent conflicts,and reduce the probability of conflicts.