Predicting System Of Asphalt Aging Based On The BP Neural Network

Under the combined effects of traffic load and climate (Oxygen, water, sunlight,etc.), Asphalt is very easy to aging, and the performance of asphalt pavement isdecreased. We learn that the factors of asphalt aging have a lot, which contains the roleof the traffic load, oxidized asphalt, rain effect, the sun’s ultraviolet radiation and thelength of time, the process has a characteristics of non-linear, chaotic nature, andlong-term memory. This paper is based on the historical data and BP neural networkmodel to research the indicators changes of asphalt aging, and establish BP networkmodel of asphalt aging to predict properties of the asphalt aging under each factors.First, three major factors that affect asphalt aging were analyzed by the means oftest analysis that asphalt-temperature, water, UV aging effects on asphalt. After acomprehensive analysis of three factors on the impact of aging derive UV> Temperature>Water. Secondly, author collects several major indicators of asphalt aging. By handlingof these aging indicators, the author can get25℃penetration,15℃ductility, softeningpoint of bitumen in different regions and time. Third, discusses the basic principles ofBP neural network、BP neural network structure, adaptability of BP neural network inthe asphalt performance prediction and BP neural network algorithm、using MATLABtool to realize the BP algorithm and so on. Fourth, making use of multi-factor(maximum temperature, minimum temperature, average annual rainfall, the annualaverage sunshine time, the age and other factors) BP neural network to predict the threeindicators to obtain the change of the three indicators of bitumen under several effectsof factors and specific network which is applicable in indicators forecast by MATLABtoolbox. Finally, by comparing the performance of highway recycled asphalt inChongqing area and network prediction, the answer has a good regularity, whichdemonstrates the accuracy of BP neural network in the forecasts.A large number of reproducible BP network experimental results show that BPneural network model can predict asphalt aging with high accuracy, this proves that theestablished methods and models employed herein is practical and effective. The Scholarscould get different times of asphalt pavement data quickly in the lab and the conclusionprovide related reference for research.