基于智能算法优化BP的航空器滑出时间预测

Prediction of Aircraft Taxi-out Time Based on Intelligent Algorithm Optimized BP

滑出时间是评估大型机场场面运行效率的主要性能指标,科学准确地预测离港航空器的滑出时间,对于提升场面运行效率至关重要。首先,分析了航空器滑出时间影响因素及相关性,构建了基于反向传播(back propagation,BP)神经网络的航空器滑出时间预测模型。针对BP神经网络存在对初始权值和阈值敏感、准确性和稳定性欠佳等缺点,分别采用粒子群优化(particle swarm optimization,PSO)算法和麻雀搜索算法(sparrow search algorithm,SSA)获取BP神经网络的最优权值和阈值,并采用中国中南某枢纽机场2周的实际运行数据对智能算法优化后的预测模型进行了验证。结果表明:滑出时间与半小时平均滑出时间、起飞队列长度、同时段滑行的离港航空器数量均有强相关性,与同时段滑入的进港航空器数量中度相关,与滑行距离和经过冲突热点区域个数相关性较弱;考虑强相关和中度相关影响因素的4元组合预测模型的预测结果最佳;智能优化算法通过获取神经网络的局部最优权重和阈值,可有效地提升航空器滑出时间预测结果的精度,但运算过程耗时也更长;基于PSO优化后的BP神经网络预测结果较优化前的平均绝对百分比误差(mean absolute percentage error,MAPE)提升了1.13%,平均绝对误差(mean absolute error,MAE)减少了4.48 s,均方根误差(root mean squared error,RMSE)减少了4.68 s;基于SSA优化后的BP神经网络预测结果较优化前的MAPE提升了3.05%,MAE减少了16.55 s,RMSE减少了14.31 s。

Taxi-out time is the main performance index to evaluate the operation efficiency of large airports.Scientifically and accurately predicting the taxi-out time of departing aircraft is very important to improve the operation efficiency.Firstly,the influencing factors and correlation of aircrafts’taxi-out time were analyzed,and the aircraft taxi-out time prediction model based on back propagation(BP)neural network was constructed.In view of the shortcomings of BP neural network,such as sensitivity to initial weight and threshold,poor accuracy and stability,particle swarm optimization(PSO)and sparrow search algorithm(SSA)were used to obtain the optimal weight and threshold of BP neural network respectively.The prediction model optimized by the intelligent algorithm was verified by using the actual operation data of a hub airport in central and southern China for two weeks.The results show that the taxi-out time is strongly correlated with the half-hour average taxi-out time,the take-off queue length and the number of departing aircraft taxiing at the same time,moderately correlated with the number of taxiing arrival aircrafts at the same time,and weakly correlated with the taxiing distance and the number of conflict hot spots.The prediction result of the 4-elements combination prediction model considering the influencing factors of strong correlation and medium correlation is the best.The intelligent optimization algorithm can effectively improve the accuracy of aircraft taxi-out time prediction by obtaining the local optimal weight and threshold of neural network,but the calculation process takes longer.The prediction result of BP neural network based on PSO optimization is 1.13%higher than that of mean absolute percentage error(MAPE)before optimization,mean absolute error(MAE)is reduced by 4.48 s and root mean squared error(RMSE)is reduced by 4.68 s.And the prediction result of BP neural network optimized based on SSA is 3.05%higher than that of MAPE before optimization,MAE is reduced by 16.55 s and RMSE is reduced by 14.31 s.