Expressway Maintenance Cost Analysis And Prediction Algorithm Model

At present,a convenient and fast expressway network has been formed in our country.In the future,the expressway maintenance task is heavy and difficult.Faced with the restriction of limited maintenance funds and technical level,maintenance has become an important aspect of expressway operation and management.Highway maintenance costs have many influencing factors and complex categories,in order to discuss the trend characteristics and influencing factors of maintenance cost and predict the maintenance cost more scientifically and effectively,this paper mainly studies the two aspects of maintenance cost characteristics and maintenance cost prediction model algorithm.This paper first analyzes the highway maintenance cost data,The daily maintenance cost of highways has a significant growth trend with the operating years,and it accounts for a large proportion of operating costs.From the micro and macro analysis,it is found that the maintenance cost and the investment amount of single kilometer,the mileage of operation,the width of the roadbed,the number of interchanges,the average distance between toll stations,the average distance between service areas,the proportion of bridge length,the proportion of tunnel length,traffic flow,passengers The goods,the average temperature,the total industrial output value above the scale and other factors are related.Using the principal component factor analysis method,the information between the factors is further explored by rotating the common factors,and the main factors affecting the maintenance cost are the bridge length,tunnel length and traffic volume.Principal component regression analysis was carried out on the results of factor analysis,and the test results showed that the regression model had better fitting and higher reliability.Secondly,based on the grey system theory and data timing characteristics,the variation law of maintenance cost with highway operation time was discussed.Through data association and mining analysis of historical cost data,outlier points were found out by data consistency method,and maintenance cost data were grouped,and a gray GM(1,1)maintenance cost prediction model was constructed.To further improve the prediction accuracy of the model,the molecular order and the iterative process were designed,and the fractional operator GM(1,1)maintenance cost prediction model was established by using the particle swarm optimization algorithm.At the same time,the causal analysis method was used to select the vehicle flow,passenger-to-goods ratio,monthly average temperature,and total industrial output value above the scale as the influencing factors of maintenance cost.The above four factors were taken as input variables,the maintenance cost was taken as the output variable,and the transfer function of the hidden layer used a strictly increasing S-type function to establish a 3-layer BP neural network prediction model.Finally,taking Shaanxi Yulin Expressway as an example,the parameters of three established maintenance cost prediction models were analyzed and applied.The results showed that the prediction accuracy of the optimized fractional operator GM(1,1)maintenance cost prediction model was higher than that of the grey GM(1,1)maintenance cost prediction model of r=1 and the maintenance cost prediction model based on BP neural network.Overall passed the residual test,the total average relative error of the model was 7.55%,and the root-mean-square error was 0.70.The forecast results can provide decision support for highway operation management and provide a scientific basis for cost control.