LAUSR

The ecological environment in high-altitude areas are fragile. Improper highway construction or excessive vehicle emission may lead to irreparable damages to the natural environment in such areas. To ensure the sustainable development of the ecological environment in high-altitude areas, it is essential to evaluate the influence of highways construction and operation on the environment in such areas. In this article, using the normalized difference vegetation index (NDVI) to indicate environmental changes in the high-altitude areas road domain. Studying sections from Pengjiazhai Town to DaYagen, DaYagen to Dongxia Town and Dongxia Township to Riyue Tibetan Town along the National Highway G214, with the altitude ranging from 2620~3760 m. Four different machine learning methods including Extreme Learning Machine, Wavelet Neural Network, BP Neural Network and Cubic Smoothing Index are then applied to analyze the NDVI changes in study areas in order to build prediction models. The results of the MAE, WMAE, RMSE and R value of the four different models show that the Wavelet Neural Network model works the best in predicting the NDVI in high-altitude areas. Based on this result, it is suggested that Wavelet Neural Network is more suitable for the intelligent prediction of road vegetation coverage in high-altitude areas.