基于Markov转移矩阵预测机场道面使用状态

Airport Pavement Performance Estimation Based on Transition Probabilities in Markov Chain

摘要:

应用Logistic回归方程建立了连续变量（道面厚度 、交通量 、道面使用时间）和分类变量（道面养护等级、道面所处的自然区域以及道面结构类型）与Markov转移概率的回归关系.选取我国多个机场实测道面状况指数数据作为基础数据，进行了模型的参数估计及显著性检验，并举例分析模型预测效果.改进了Markov概率预测模型中的参数标定方法，提高了预测精度，并在一定程度上解决了因为机场道面观测数据少，传统预测模型无法获得稳定参数估计的问题，拓展了模型的适用范围.

Abstract:

The logistic regression equation was used to establish regression relationship among continuous variables such as thickness, traffic level, pavement age, categorical variables such as maintenance grade, climate sector, pavement structure types and the Markov probability prediction. Some airports’ actual measurement pavement condition index (PCI) data were used as data source in the parameter estimation and significance test of the model. The parameter calibration method of Markov probability prediction model, and the accuracy of the predictions were improved. The scope of the application of model was expanded and the problems of insufficient observation data and unstable parameter estimation using the traditional forecasting model were resolved.