Recent observations of global warming phenomena have necessitated the evaluation of the service performance of asphalt pavements, which is substantially influenced by surface temperature levels. This study employed twelve distinct machine learning algorithms—K-neighbors, linear regression, multi-layer perceptron, lasso, ridge, support vector regression, decision tree, AdaBoost, random forest, extra tree, gradient boosting, and XGBoost—to predict the surface temperature of asphalt pavements. Data were sourced from the Road Weather Information System of Iowa State University, comprising 12,581 data points including air temperature, dew point temperature, wind speed, wind direction, wind gust, and pavement sensor temperature. These data were segmented into training (80%) and testing (20%) datasets. Analysis of model outcomes indicated that the Extra Tree algorithm was superior, exhibiting the highest R$^2$ value of 0.95, whereas the Support Vector Regression algorithm recorded the lowest, with an R$^2$ value of 0.70. Furthermore, Shapley Additive Explanations were utilized to interpret model results, providing insights into the contributions of various predictors to model outcomes. The findings affirm that machine learning algorithms are effective for predicting asphalt pavement surface temperatures, thereby supporting pavement management systems in adapting to changing environmental conditions.

To address the lack of clear formulae for calculating the circumferential stress in steel epoxy sleeve-reinforced pipelines under internal pressure, this study constructs a mechanical model based on the specific stress characteristics of these pipelines. Using stress solution methods and deformation compatibility relationships, theoretical formulas for circumferential stress in the pipeline layer, epoxy resin layer, and sleeve layer under internal pressure are derived. The theoretical formulas are validated through numerical simulations using ANSYS software, which includes models with and without flanges. The calculations were performed for common pipelines with outer diameters of 219mm, 660mm, and 1219mm. The results show that the discrepancies between theoretical and numerical solutions of circumferential stress in all layers of both model types are within 10%. Specifically, the circumferential stress in the pipeline layer of the flanged model is lower than that of the non-flanged model and also lower than the theoretical values. The error between the theoretical and numerical solutions for pipelines of different diameters does not exceed 10%, confirming the validity and applicability of the theoretical formulas. This suggests that using the simplified mechanical model for circumferential stress calculations ensures a conservative approach for the structural assessment of pipelines. The formulas provided herein can serve as a reference for the design and evaluation of steel epoxy sleeve-reinforced pipelines under internal pressure.