【摘要】 为提高铁路客运量预测精度，提出Newton插值法对客运量原始数据进行预处理以解决因节假日或重大事件造成的数据异常问题。另外，引入超松弛技术(OR)对铁路客运量预测结果进行修正，提出非线性递减权重改进粒子群算法以优化松弛因子。最后，将Newton插值法、超松弛技术与GM(1,1)和BP神经网络预测相结合，提出铁路客运量Newton-GM-BP-OR组合预测方法，并以北京市铁路客运量预测为例验证预测方法的有效性。研究结果表明，基于Newton插值法处理异常客运量数据的预测效果较基于原始数据序列更好，改进的粒子群算法在求解松弛因子过程中体现出更好的寻优能力和收敛速度，且超松弛技术对GM(1,1)和BP神经网络预测结果的修正也使得Newton-GM-BP-OR组合预测方法具有更高的预测精度。更多还原

【Abstract】 In order to improve the prediction accuracy of railway passenger volume, the Newton interpolation method was proposed to preprocess the raw data on passenger volume to solve the problem of abnormal data caused by holidays or major events. In addition, the over relaxation technology(OR) was introduced to correct the prediction results of railway passenger volume, and the improved the particle swarm optimization algorithm of non-linear decreasing weight proposed to optimize the relaxation factor. Finally, in combination with the Newton interpolation method, over-relaxation technology and GM(1, 1) and BP neural network prediction, a Newton-GM-BP-OR combined prediction method for railway passenger traffic was proposed, and the validity of the prediction method verified by taking Beijing railway passenger volume prediction as an example. The research results show that the prediction effect of abnormal passenger traffic data processing based on Newton interpolation is better than that based on the raw data sequence, and the improved particle swarm algorithm shows better optimization ability and convergence speed in solving the relaxation factor. Moreover, the correction of the prediction results of GM(1, 1) and BP neural network by the over relaxation technology also enables higher prediction accuracy of the Newton-GM-BP-OR combined prediction method.更多还原