摘要
通过现场沉降监测,获取了高速铁路桥梁桩基沉降的大量实测数据。据此分析了成桥过程中桩基沉降发展的规律及其沉降曲线特点。对桩基最终沉降,采用单项预测模型进行了拟合预测,通过预测效果的评价指标分析,指出了客运专线铁路无碴轨道铺设条件评估技术指南中评价预测模型合理性仅局限于相关系数指标的不足,建议引入均方误差MSE、平均绝对百分比误差MAPE和平均绝对误差MAE指标进行综合检验、评价,并提出了这些评价指标相应的参考值。引入组合预测模型进行高速铁路桥梁工后沉降预测,对比分析现有组合预测模型的特点,提出了高速铁路桥梁桩基工后沉降预测应遵循的具体步骤和原则。研制了桥梁桩基沉降加权组合预测电算程序BriFSCF,可大大提高预测准确度和效率。通过工程实例验证表明,最小二乘准则下的最优组合预测方法效果较好,可作为预测的优选模型。其结果可供高速铁路合理铺设轨道时间的确定、工后沉降计算参考和利用。
A large number of accurate and reliable settlement data were obtained via on-site long-term monitoring test.The basic settlement law of pile foundation and characteristics of settlement curve were studied.Single forecast model was used to fit the monitoring data.The inadequacy that the evaluation of reasonable prediction model was limited to the correlation coefficient index in current guidebook was pointed out.It was proposed that mean-square error,mean absolute percentage error and average absolute error were introduced to comprehensively examine,evaluate the fitting effect and the reference value corresponding to the evaluation index was recommended.Combination forecast model for high-speed railway bridge post-construction settlement prediction was proposed;the characteristics of existing combination prediction models were analyzed and compared;specific steps and basic principles of bridge pile foundation post-construction settlement forecast were brought forward;and corresponding computer program was developed to improve forecast accuracy and efficiency.It was found that,by engineering examples,the optimal combination forecast method based on least-squares criterion could be thought of as the favor model owing to effectively and comprehensively improving the prediction accuracy.The valuable supplement and reference were provided to confirm reasonable time to construct railway and calculate post-construction settlement.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2011年第11期3341-3348,共8页
Rock and Soil Mechanics
基金
国家自然科学基金资助项目(No.50678175)
铁道部科技研究开发计划项目(No.2005K002-A-1)
关键词
高速铁路
桥梁桩基
现场监测
工后沉降
组合预测模型
high-speed railway
bridge pile foundation
on-site monitoring
post-construction settlement
combination forecast model