類神經網路於飛航網路運量預測之應用

Air Transport Demand Forecasting in Routes Network by Artificial Neural Networks

郭仕堯(Shih-Yao Kuo) ； 蕭樂群(Le-Chung Shiau) ； 張原賓(Yuan-Pin Chang)

《航空太空及民航學刊.系列B》 42卷1期 (2010/04) Pp. 67-72

https://doi.org/10.6124/JAAA.201004_42(1).08

摘要

航空產業從策略規劃到實際營運，運量需求預測都是不容輕忽的重要基礎，常見的方法有時間序列分析、重力模式、灰色理論與類神經網路。本研究使用類神經網路建立多輸入、多輸出之數學模式，與傳統單輸入、單輸出之時間序列分析、灰色理論不同，亦不同於傳統類神經網路?單輸出，僅考慮單一航站或單一航線之模式。本研究可補強時間序列分析與灰色理論不足之處，亦可改善重力模式必須事先確定數學方程式形式與參數之缺點，實證結果指出此模式能準確預測飛航網路運量。

關鍵字

運量預測 ； 飛航網路 ； 類神經網路

並列摘要

Aviation industry relies strongly on air transport demand forecasting for developing the operation strategy. Time series analysis, gravity model, grey theory and artificial neural networks are familiar tools for forecasting air traffic. In this article, artificial neural networks were employed to establish mathematical model with multiple inputs and multiple outputs, which is different from time series analysis and grey theory considered only single input and single output. The traditional analyses for air transport demand forecasting by artificial neural networks consider single output, as single airport or one route is considered in general. This research overcomes the shortcomings of time series analysis and grey theory. It also improves the weakness of gravity model that must confirm the explicit equation in advance. The results indicate that the novel model may accurately forecast the air transport demand in routes network.

並列關鍵字

Air transport demand forecasting ； Routes network ； Artificial neural networks