Volume 3 · Issue 8 (2026)
DOI number:
10.66521/2938-9933-2026081301
Adhesion Prediction Model for Reclaimed Asphalt Mixture Based on Machine Learning
Yan Zhang*, Yiting Li
Dalian University of Engineering, Dalian 116600, China
Corresponding Author: Yan Zhang (861257865@qq.com)
Abstract: The interfacial adhesion between reclaimed asphalt and aggregates is a critical factor determining the durability of the mixture, especially when a high proportion of reclaimed asphalt mixture is used. However, existing research is mostly limited to macroscopic tests and empirical models, lacking systematic investigation into the structure-property relationship between the microscopic composition of the asphalt system and its adhesion performance. In this study, a prediction model for reclaimed asphalt adhesion was constructed through multi-scale characterization techniques and based on machine learning. Convolutional Neural Network (CNN) and Elman Neural Network were selected for training, and the accuracy was evaluated using Mean Absolute Error (MAE), Relative Percentage Error (RPE), and coefficient of determination (R^2). Finally, the Elman Neural Network was selected as the reclaimed asphalt adhesion prediction model through a multi-index weighted scoring method, and adhesion evaluation indexes based on surface free energy and AFM force curves were preferred to construct a multi-scale adhesion evaluation parameter system. The results show that the Elman prediction model constructed in this paper has a relative percentage error RPE value of less than 1, a mean absolute error MAE value of less than 5, and a coefficient of determination R^2 value greater than 0.9, which can provide a reliable basis for RAP content optimization and further promote the application and development of reclaimed asphalt technology in engineering practice.
Keywords: Reclaimed Asphalt, Adhesion, Microscopic Composition Structure, Machine Learning, Elman Neural Network
References
[1]
Luo Zhongjian, Zhu Chenxu, Han Zhongming, et al. Research on Road Performance of High-Content RAP Composite Recycled Asphalt Mixture [J/OL]. Plastics Science and Technology, 2025,(10):60-65.
[2]
Huang You, Feng Boxiong, Liu Zhaohui, et al. Research on High RAP Content and High Modulus Asphalt Mixture [J]. China and Foreign Highways, 2025,45(05):131-140.
[3]
Yu Xin, Gao Yuchao, Zhou Jie. Research and Application Progress of RAP plant-mixed hot Regeneration Based on oil-stone separation [J]. China and Foreign Highways, 2025,45(05):1-18.
[4]
Qin Binbin. Performance Evaluation of Plant-mixed Hot Recycled Asphalt Mixture Based on RAP and Waste PET [J]. Transportation Energy Conservation and Environmental Protection, 2025,21(05):85-89.
[5]
Chen Yib, Li Feng, Zhou Siqi, et al. In-situ density prediction Model of asphalt pavement Based on Machine Learning Algorithm [J]. China Journal of Highway and Transport,2023,36(12):333-344.
[6]
Cheng Shengjie. Research on Prediction of Adhesion Characteristics of Warm-Mixed Recycled Asphalt - Aggregates Based on Machine Learning [D] Chongqing Jiaotong University, 2025.
[7]
Shen Qiu, Xiao Peng, Gu Wan, et al. Research on Performance Prediction of Recycled Asphalt Mixture Based on Genetic Neural Network [J]. Highway Engineering, 2020,45(02):61-67.
[8]
Tang Liangqi. Research on Rapid Prediction Method of Asphalt Quality Recognition and Performance Based on Machine Learning [D] Guangdong University of Technology, 2025.
[9]
Sheng Jiahao, Liu Li, Liu Zhaohui, et al. Asphalt Mixture Performance Prediction Method Based on Genetic Algorithm Optimized BP Neural Network [J] Science Technology and Engineering, 2025,25(03):1214-1224.
[10]
Bishop C M. Model-based machine learning [J]. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2013, 371(1984): 20120222.
[11]
Liang Z, Wang Q. A robust model averaging approach for partially linear models with responses missing at random[J]. Scandinavian Journal of Statistics, 2023, 50(4): 1933-1952.
[12]
Wu C, Zhang Q, Cheng Y, et al. Novel three-way generative classifier with weighted scoring distribution[J]. Information Sciences, 2021, 579: 732-750.
Download PDF