Abstract
Mastering the creep characteristics of artificial frozen soil and scientifically evaluating the creep model is an important guarantee for the safety of subway tunnel freezing construction. Base on the construction of Nantong metro tunnel, the uniaxial compressive strength tests of the artificially frozen soft soil were carried out to obtain the influence law of temperature on the uniaxial compressive strength, and the uniaxial creep tests were carried out to obtain the influence law of temperature and stress grade on creep, at - 5, - 10 and - 15 °C. The experimental results show that the creep characteristics of frozen soft soil specimens have obvious fuzzy randomness. The traditional ant colony algorithm is improved by optimizing the pheromone fuzzification coefficient, which improves the search efficiency and avoids the local optimum effectively. Subsequently, the improved fuzzy ant colony algorithm is used to invert the flexibility parameters of commonly used permafrost creep models. The fuzzy weight of evaluation index and the fuzzy random evaluation matrix were determined to evaluate the optimal creep model under three different stress levels of frozen soft soil. Finally, the reliability of the fuzzy random evaluation method was verified by engineering measured data.