Abstract:

Improves the precision of the segment demand forecast is the important research topic in aviation revenue management system．Used the EM algorithm[2] to restore the historical demand which had constained by resource，then obtain the more precise demand data．The tickets of the scheduled flight can book advance．So centralized the segment’s historical booking data into a booking matrix for using more known information and decrease the amount of the data to forecast．Then used the Winters model to forecast the segment demand．Calculated the data within the segment from Chengdu to Gangzhou，the average error of the experimental results is 4.9 percents，and the average error of the Winters model is 21.7 percents，and the feasibility and effectiveness of the proposed method has been illustrated by the experiment．

Key words: airline demand forecasting, winters model, EM algorithm, booking matrix

摘要： 提高航线需求预测精度是航空客运收益管理系统的重要研究课题。提出一种首先运用EM算法对航线历史订座数据中由于受资源限制而没有被满足的需求进行修复，取得更为准确的需求数据，然后，针对航班销售具有提前预订的特点，将航线历史订座数据整合成一个提前日矩阵，使预测时能够利用更多的已知信息并减少了计算数据量，最后再运用Winters模型进行预测的航线需求预测方法。以成都到广州航线历史数据为例仿真计算，预测结果的平均误差为4.9％ ，而直接采用winters法进行预测的平均误差为21.7％ ，验证了算法的可行性和有效性。

关键词: 航线需求预测, Winters模型, EM算法, 提前日矩阵