基于基因表达式算法的路面材料性能预测

发布时间：2018-08-17 12:38

【摘要】：作为路面结构设计和施工的重要依据,路面材料性能的研究一直是道路工程方向研究的热点。本文通过基因表达式编程算法对沥青混合料动态模量、再生沥青混合料动态模量、混凝土自收缩等路面材料性能预测模型进行研究,所得到的分析成果对道路工程、结构工程等领域具有重要的实际意义。本文首先系统分析了沥青混合料动模量、再生沥青混合料动模量、混凝土自收缩的研究现状。然后,借鉴前人的研究成果,应用基因表达式编程算法基本理论,对沥青混合料动模量、再生沥青混合料动模量、混凝土自收缩进行了比较深入的分析研究。主要研究内容可总结如下几点:(1)采用基因表达式编程算法进行沥青混合料动态模量的预测,以沥青混合料动态模量的八个主要影响因素:沥青混合料的空隙率(aV),有效沥青含量(b effV),沥青黏度(η),荷载频率(f),沥青混合料在19.0、9.5、4.75mm号筛筛后的累计筛余质量分数(34?、38?、4?),及沥青混合料在0.075mm号筛筛孔上的通过率(200?),构成预测沥青混合料动态模量模型的主要参数,通过对八个离散的参数建立基于基因表达式编程算法的沥青混合料动态模量预测模型。结果表明:预测模型得到的动态模量预测值与实测值之间具有较高的相关性;并将预测模型与Witczak1999年函数模型,韩国动态模量预测模型以及人工神经网络模型进行比较分析,结果表明基因表达式编程算法预测沥青混合料动态模量具有简单可靠的优点。(2)应用基因表达式编程算法研究再生沥青混合料动态模量预测模型,基于热拌沥青混合料动模量的研究成果,参照采用Witczak 1999沥青混合料动模量预测模型中8个影响因素,另外增加回收沥青路面混合料的掺配比,作为再生沥青混合料动模量的9个输入参数。采用基因表达式编程算法对9个影响因素离散分析得到再生沥青混合料动态模量预测模型。分析动模量预测值与实测值的拟合度以及动态模量与掺配比之间的相关性。并对各影响因素与动模量之间的敏感性进行了分析。结果表明:预测模型得到的预测值与实测值之间具有较好的拟合度,对再生沥青混合料动态模量的研究具有一定的参考价值。(3)进行混凝土自收缩试验,分析不同水灰比、硅粉掺加量对自收缩的影响。运用基因表达式编程算法研究混凝土自收缩与其主要影响因素(水灰比、矿物掺合料、骨料含量、水泥浆体含量、高效减水剂、养护温度、养护龄期)之间的关系,得到混凝土自收缩预测模型,并将自收缩预测值与实测值进行拟合度分析,在此基础上,分析各主要影响因素与自收缩之间的相关性。研究表明:基因表达式编程算法的混凝土自收缩预测模型在精度上符合工程设计要求,对混凝土结构设计有一定的指导意义。最后,在全面总结全文工作的基础上,对进一步的研究工作提出了一些建议与展望。
[Abstract]:As an important basis for pavement structure design and construction, pavement material performance research has always been a hot topic in the direction of road engineering. In this paper, the dynamic modulus of asphalt mixture, dynamic modulus of recycled asphalt mixture, autogenous shrinkage of concrete and other pavement material performance prediction models are studied by genetic expression programming algorithm. The analysis results have important practical significance for road engineering, structural engineering and other fields. Firstly, this paper systematically analyzes the research status of dynamic modulus of asphalt mixture, dynamic modulus of recycled asphalt mixture and autogenous shrinkage of concrete. The main research contents can be summarized as follows: (1) The genetic expression programming algorithm is used to predict the dynamic modulus of asphalt mixture, and the eight main factors affecting the dynamic modulus of asphalt mixture are: the void fraction (aV) of asphalt mixture. Available bitumen content (b effV), bitumen viscosity (_), loading frequency (f), cumulative residual mass fraction (34?, 38?, 4?) of bitumen mixture after 19.0, 9.5, 4.75 mm sieve, and the throughput of bitumen mixture on 0.075 mm sieve hole (200?) constitute the main parameters of predicting the dynamic modulus model of bitumen mixture. Through eight discrete parameters A prediction model of dynamic modulus of asphalt mixture based on genetic expression programming algorithm is established.The results show that the predicted values of dynamic modulus obtained by the prediction model are highly correlated with the measured values.The prediction model is compared with Witczak 1999 function model,Korea dynamic modulus prediction model and artificial neural network model. The results show that the genetic expression programming algorithm is simple and reliable in predicting the dynamic modulus of asphalt mixture. (2) Using the genetic expression programming algorithm to study the dynamic modulus prediction model of recycled asphalt mixture. Eight influencing factors in the prediction model are added to the mixture ratio of recycled asphalt pavement mixture as nine input parameters of the dynamic modulus of recycled asphalt mixture. The results show that the predicted value of the predicted model has a good fit with the measured value, and it has a certain reference value for the study of the dynamic modulus of recycled asphalt mixture. (3) Concrete. The autogenous shrinkage test was carried out to analyze the effect of different water cement ratio and silica fume content on the autogenous shrinkage of concrete.The relationship between autogenous shrinkage of concrete and its main influencing factors (water cement ratio, mineral admixture, aggregate content, cement paste content, superplasticizer, curing temperature and curing age) was studied by genetic expression programming. The results show that the prediction model of concrete autogenous shrinkage based on genetic expression programming meets the requirements of engineering design in precision and has certain guidance for concrete structure design. Finally, on the basis of a comprehensive summary of the full-text work, some suggestions and prospects for further research are put forward.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U414

【参考文献】