In this thorough examination, we dive deep into the long-lasting mechanical characteristics of fine-grained saline soil subgrades, aspiring to establish a precise and reliable collection of predictive models. Our objective is to provide a solid scientific footing for the design and ongoing upkeep of road networks within saline soil environments. Analyzing prolonged monitoring data across diverse highway subgrades within a prototypical saline soil locale, we unveil the intricate temporal fluctuations and environmental sensitivities of the soil’s mechanical properties under continuous load. Precisely, the subgrade’s compressive modulus dwindled by 15%, while shear strength declined by 8% over a five-year period. These trends intensify during rainy and scorching seasons, with drops surpassing 20% and 12% respectively. Leveraging this intricate data, we deploy nonlinear regression analysis and sophisticated machine learning algorithms to construct a predictive model tailored for the long-term mechanical properties of fine-grained saline soil roadbeds. This model integrates a multitude of factors, including load duration, temperature, humidity, and more, delivering accurate forecasts of key subgrade indicators like compressive modulus, shear strength, and beyond. In the verification stage, compared with the measured data, the error rate of the model prediction results is controlled within 5%, showing high prediction accuracy and stability. In addition, we also carried on the sensitivity analysis to the model, found that the load size and the duration of the impact on the mechanical properties of the roadbed is the most significant. Therefore, in the design of road engineering in saline soil areas, the influence of these factors should be fully considered, and reasonable engineering measures should be taken to ensure the safety and durability of roads. This study not only provides effective data support for the long-term mechanical performance evaluation of fine-grained saline soil roadbed, but also provides an important theoretical reference for engineering practice in related fields.
Read full abstract