公路养护成本预测的智能算法分析研究

发布时间：2018-04-18 01:24

本文选题：支持向量机 + 布谷鸟搜索算法　；参考：《武汉理工大学》2015年硕士论文

【摘要】：随着我国越来越多的高速公路投入运营,其养护水平的高低直接影响到高速公路的使用寿命和通行质量,我国已经开始进行高速公路养护市场化的改革。及时有效地进行小修保养,可以减少进行大中修的次数,延长公路的使用寿命。对于小修保养工程量影响因素及其预测的研究,能够为养护部门在进行公开招标时提供数据支撑和决策支持,更加合理地分配有限的养护资金,让养护资金得到最大化的利用,是具有重要意义的。本文通过调查研究国内外相关资料的基础上,结合当前我国高速公路养护的特点和现状,概括总结了高速公路沥青路面小修保养路面破损维修量的影响因素,并在此基础之上,引入支持向量机回归对路面破损维修工程量进行预测。本文将一种新型元启发式算法——布谷鸟搜索算法应用于支持向量机参数寻优中,与原有的基于遗传算法优化支持向量机参数的方法进行对比,表明了利用布谷鸟搜索算法优化支持向量机参数的可行性以及所具有的优势。在此基础上对布谷鸟搜索算法进行改进,提出了基于高斯扰动和基于梯度下降法的布谷鸟搜索算法优化支持向量机参数,与原有方法进行比较,分别使用普通布谷鸟搜索算法、基于高斯扰动的布谷鸟搜索算法和基于梯度下降法的布谷鸟搜索算法优化支持向量机参数后对路面破损维修工程量进行预测,实验结果表明,采用基于梯度下降法的布谷鸟搜索算法优化支持向量机参数的预测结果相对误差为9.75%,3-折交叉验证的均方误差为170.7395,参数寻优的迭代次数为130次,优于普通布谷鸟搜索算法优化支持向量机参数的198次迭代和基于高斯扰动的布谷鸟搜索算法优化支持向量机参数的146次迭代,且在这三种方法中,其收敛下降的速度是最快的。
[Abstract]:With more and more expressways put into operation in our country, the level of maintenance has a direct impact on the service life and traffic quality of freeway, so our country has begun to carry out the reform of freeway maintenance marketization.Timely and effective minor repair and maintenance can reduce the number of major repairs and prolong the service life of the highway.The research on the factors affecting the quantity of minor repairs and maintenance and its prediction can provide data support and decision support for the maintenance department in conducting public bidding, allocate the limited maintenance funds more reasonably, and maximize the utilization of the maintenance funds.Is of great significance.On the basis of investigation and study of relevant data at home and abroad, combined with the characteristics and present situation of expressway maintenance in China, this paper summarizes the factors affecting the amount of damaged pavement maintenance of highway asphalt pavement, and on this basis,Support vector machine regression is introduced to predict the amount of pavement damage and maintenance.In this paper, a new meta-heuristic algorithm, Cuckoo search algorithm, is applied to the parameter optimization of support vector machine, which is compared with the original method based on genetic algorithm to optimize the parameters of support vector machine.The feasibility and advantages of using cuckoo search algorithm to optimize support vector machine parameters are demonstrated.On this basis, an improved Cuckoo search algorithm is proposed, which is based on Gao Si perturbation and gradient descent algorithm to optimize support vector machine parameters. Compared with the original method, the common cuckoo search algorithm is used respectively.After optimizing the parameters of support vector machine (SVM), the Cuckoo search algorithm based on Gao Si disturbance and the Cuckoo search algorithm based on gradient descent method are used to predict the amount of pavement damage and maintenance. The experimental results show that,Based on the gradient descent method, the relative error of the prediction result of optimizing the parameters of support vector machine by Cuckoo search algorithm based on gradient descent method is 9.75 and the mean square error of cross-validation is 170.7395, and the number of iterations of parameter optimization is 130.It is better than the 198 iterations of the ordinary cuckoo search algorithm to optimize the parameters of the support vector machine and the 146 iterations of the optimization of the parameters of the support vector machine based on Gao Si's perturbation, and the speed of convergence of these three methods is the fastest.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U418.2

【参考文献】