Research Of Maintenance Amount Prediction On The Highway With Intelligent Algorithm

More and more highways have come into service in our country. The service life and quanlity of highway is affected by the level of highway maintenance directly.The marketization reform of highway maintenance has already been begun. With routine maintenance done effectively and timely, it can decrease the times of intermediate and heavy maintenances, and extend the service life of the highway.That, the research of the affecting factors and account prediction on routine maintenance, can provide data and decision support for maintenance departments when they make an open biding.It also can make full use of limited maintenance funds.In this thesis, it makes an investigation of related data at home and broad.And according to the characteristics and current situation of the highway maintenance on our country, it gets the influence factors of highway asphalt pavement damage.Based on these factors, we predict the maintenance quantities of pavement damage with support vector machine.In this thesis, it tries to apply a new metaheuristic algorithm, cuckoo search algorithm, to optimize the parameters of support vector machine. Compared to the genetic algorithm optimization of support vector machine parameters, it shows that it is feasible and has its advantages.And then, in this thesis, two imporved algorithms are verified, cuckoo search algorithm based on gauss perturbation and cuckoo search based on gradient descent to optimize the parameters of support vector machine.we predict the maintenance quantities of pavement damage with these three methods. Experimental results show that the relative error of predicted results with cuckoo search based on gradient descent to optimize the parameters of support vector machine is 9.75%. The mean square error of 3-fold cross-validation is 170.7395 and the number of iterations is 130 times. Compared to the number of iterations with original cuckoo search,198 times,and the number of iterations with cuckoo search algorithm based on gauss derturbation,146 times, it is the best.At the same time,its convergence speed is also the fastest.