摘要
为了解决国际平整度指数IRI预测模型准确性不高的问题,以京沪高速公路实测IRI数据为基础,对log istic回归、多元回归、时间序列这3种建模方法分别进行分析.并根据京沪高速公路平整度实测数据,建立了几个有不同数量滞后值的时间序列路面平整度预测模型,根据与实测值的比较,找出最优的时间序列路面平整度预测模型.分析结果表明:利用传统的log istic回归和多元回归方法难以建立准确预测路面平整度发展趋势的模型;时间序列方法具有较高的预测精度,且其易修正性是其他预测方法所不具备的.
Prediction model of international roughness index （IRI） has a disadvantage of poor precision. Based on the IRI data of Jinghu freeway, three kinds of IRI prediction methods are analyzed in the paper： logistic regression method, multi-regression method, and time series method. With the IRI of Jinghu freeway, a time series prediction model of IRI with different number lag values is established, and by the comparison with actual IRI value, the best prediction model, time series prediction model of IRI is found. The result shows that： logistic regression model and multiegression model can not work well to predict the trend of IRI; time series model of IRI can predict the trend of IRI very well, and its easiness of correction is unique.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第4期634-637,共4页
Journal of Southeast University：Natural Science Edition
