Abstract: The transportation capacity of urban agglomerations is the strategic cornerstone of building a national comprehensive three-dimensional transportation net. In view of the fact that traditional forecasting methods were difficult to adapt to many influencing factors and had the characteristics of time-varying, coupling and strong uncertainty, a gray model-generalized regression neural network compound model was proposed to predict the transportation capacity of urban agglomerations in the long-term. Firstly, the LASSO method was used to screen out the main influencing factors to reduce data complexity. GM (1,1) model was used to weaken the randomness of data series, predict the change trend of time series of influencing variables and fill in the lack of data. Then, GRNN model was trained by the dataset of Beijing-Tianjin-Hebei urban agglomeration from 2000 to 2019. According to the influencing factors predicted by GM (1,1) from 2020 to 2025, the dynamic trend of transportation capacity in the future years was obtained. The results show that the accuracy of the compound prediction model is better than that of traditional methods, which effectively reduces the uncertainty of the small sample prediction. Finally, combined with prediction results, the development direction of the core location cities of Beijing-Tianjin-Hebei urban agglomeration is analyzed, which makes a forward-looking discussion in order to help build a new development pattern with urban agglomeration as an important starting point.

Key words: traffic and transportation engineering; urban agglomeration; gray model-generalized regression neural network model; transportation capacity prediction

摘要： 城市群交通运输能力是构建国家综合立体交通网的战略基石。鉴于传统预测方法难以适应城市群交通运输能力影响因素众多且存在时变、耦合、不确定性强等特征，提出了一种灰色-广义回归神经网络的复合模型，以预测未来城市群交通运输能力。首先，选用LASSO算法筛选主要影响变量来降低数据复杂度，运用GM(1,1)模型弱化数据序列的随机性，预测影响变量时间序列的变化趋势，并填补数据缺失。然后，以2000—2019年京津冀城市群的数据集训练GRNN模型，根据GM(1,1)模型预测出的2020—2025年城市群交通运输能力影响因素，得出未来年份交通运输能力动态趋势。结果表明，复合预测模型精度优于传统方法，有效减少了小样本预测的不确定性。最后，结合预测结果分析了京津冀城市群核心区位城市的发展方向，为助力构建以城市群为重要抓手的新发展格局进行了前瞻性探讨。

关键词: 交通运输工程；城市群；灰色-广义回归神经网络模型；交通运输能力预测