Research on carbon emission forecasting during expressway maintenance period based on GWO-Markov optimized grey model.

To address the problems of limited carbon emission data samples from highway maintenance and insufficient accuracy of traditional prediction models, this study adopts the Grey Wolf Optimizer (GWO) to optimize the Grey Model GM(1,1) and combines the Markov Chain for residual correction, thereby constructing the GWO-Markov-GM (GMG) model. The carbon emission data of maintenance projects from 2015 to 2021 of 4 expressway in Anhui Province were calculated using the carbon emission factor method combined with pricing documents. Based on this data, the GMG model was trained and compared with traditional models such as GM and PSO-GM (Particle Swarm Optimization-Based Grey Model). Meanwhile, comparative experiments with non-grey benchmark models including Multiple Linear Regression (MLR), Autoregressive Integrated Moving Average (ARIMA), and Random Forest (RF) demonstrate that the model still exhibits significant comprehensive advantages in small-sample carbon emission prediction tasks. The results show that the Mean Absolute Error (MAE) of the GMG model is 26.006, which is significantly lower than that of the comparison models. The Root Mean Square Error (RMSE) is reduced by 47.913 compared with the average value of the comparison models, the Mean Absolute Percentage Error (MAPE) is reduced by 18.545%, the Coefficient of Determination (R2) reaches 0.973, the Posteriori Error Ratio (C) is 0.139, and the Small Error Probability (P) is 1. The GMG model has significant advantages in prediction accuracy, stability, and data fitting ability. Based on this, the carbon emissions of these 4 expressway from 2022 to 2024 were predicted, providing quantitative support for the green development in this field.