Research On Construction Quality Control Of Graded Crushed Stone Of Inverted Structure Asphalt Pavement Based On Artificial Neural Network

In recent years,more and more examples have proved that the semi-rigid base asphalt pavement structure easy to form reflective cracks objectively.And once cracks-damage occurs,the maintenance of the pavement structure is often faced with high energy consumption and high cost.Inverted Asphalt Pavement has excellent anti-reflection crack and certain drainage performance because of the organic combination of graded crushed stone,asphalt stabilized gravel and semi-rigid material,and has good economic benefits.The construction quality of graded gravel layer of inverted structure asphalt pavement is an important factor to ensure that the pavement structure has good mechanical properties and stability.So strengthening the construction quality control of graded gravel layer is the key guarantee to improve the construction quality of the whole inverted structure asphalt pavement.The construction quality control of graded gravel is a nonlinear comprehensive problem with many influencing factors.At present,the construction quality control mainly focuses on the post control.This paper introduces the artificial neural network to establish the construction quality control and prediction model of graded gravel,simulates the experts to predict the quality level of graded gravel layer after paving,and then judges whether the current construction quality control measures are effective and how to improve according to the prediction results.In this paper,the characteristics and the key points of construction quality control of graded gravel for inverted structure asphalt pavement are analyzed from the whole construction process,and the quality control index system of graded gravel construction is established.In order to eliminate the correlation between the indexes,6 principal components were extracted by using principal component analysis as the input of neural network model.Then based on BP neural network and MATLAB platform,the construction quality prediction and evaluation model of graded crushed stone base is established,The data samples of 32 gradated gravel construction projects collected in Guangdong were used for training and testing.,and verified with engineering examples.A referable and practical control method is provided for the construction quality management of graded crushed stone base on inverted pavement.Finally,the paper further compares the errors of the construction quality control and prediction models of graded crushed stone base based on BP neural network,RBF neural network and GR neural network respectively,which provides a train of thought for the selection of the construction quality control neural network model of graded crushed stone base.