基于动态GM-Markov优化模型的高铁路基冻胀变形预测研究

由于传统灰色模型在预测波动性较大的数据时精度不高，提出一种改进的动态GM-Markov预测模型。利用非等间距加权矩阵与无偏优化对灰色模型进行改进，通过原始序列的动态更新实现模型的参数更新，在此基础上利用马尔科夫模型进行残差修正，得到改进动态GM-Markov预测模型。利用某高铁路基冻胀变形监测数据进行实例分析，将改进的动态GM-Markov模型预测结果与灰色以及非等间距无偏灰色模型、最优组合模型预测结果进行对比分析，结果表明：改进的动态GM-Markov模型对于波动性较大的冻胀变形可以取得较好预测效果，提高了预测精度与稳定性。

Prediction of Frost Heave Deformation of High-speed Railway Subgrade Based on Dynamic GM-Markov Optimization Model

Because the accuracy of the traditional grey model is not high in predicting the data with high fluctuation, an improved dynamic GM-Markov prediction model has been proposed. The grey model has been improved by non-equal spacing weighted matrix and unbiased optimization, and the parameters of the model are updated by dynamic updating of original sequence. On this basis, the improved dynamic GM-Markov prediction model is obtained by residual correction of Markov model. Based on the monitoring data of the frost heaving deformation of a high-speed railway foundation, the improved dynamic GM-Markov model prediction results has been compared with the prediction results of grey model, non-equidistance unbiased grey model and optimal combined model. The results show that the improved dynamic GM-Markov model can achieve better prediction effect for frost heave deformation with large fluctuation and improve the prediction accuracy and stability.