关键词: 时间序列聚类, OD 需求预测, WOA-BP 神经网络, 鲸鱼优化算法

Abstract: Inter-airport OD demand forecasting plays a crucial role in route analysis and sales decision. Based on 2,143 in鄄 ter-airport OD demand time series from China, this paper proposes a WOA-BP neural network model that inte鄄 grates the whale optimization algorithm (WOA) with a back-propagation (BP) neural network. Through trend clus鄄 tering and step size optimization, the model automatically adapts to different sequence trend patterns under a u鄄 nified training process. The experimental results show that the mean absolute percentage error of the WOA-BP neural network is reduced by 8.0%, the correlation coefficient is improved by 2.1%, and the mean square error is reduced by 82.9% compared with the traditional BP neural network. The model shows good robustness and gen鄄 eralization ability in several typical inter-airport OD demand forecasts, which can significantly

Key words: time series clustering, OD demand forecasting, WOA-BP neural network, whale optimization algorithm