In the context of the “carbon peaking and carbon neutrality” goals

the specific heat capacity of soil

as one of the important thermophysical parameters

plays an important role in many fields such as energy geo-structure

subway ventilation calculation

buried pipelines

nuclear waste disposal

artificial ground freezing

agricultural irrigation

etc. A specific heat capacity test was carried out on clay samples with different initial dry densities and initial water contents under different temperature conditions

and the variation patterns and mechanisms were investigated. The test results indicate that the specific heat capacity of clay increases linearly with the increase of initial dry density and initial water content

and is proportional to the third power of temperature. The mechanism of the variation of specific heat capacity is related to the variation of three-phase percentage

solid-phase lattice vibration

and water-ice phase transition. A model for predicting the specific heat capacity of clay considering multi-factors is established and cross-validated with back propagation artificial neural network (BP-ANN) model

simplified theoretical model and other empirical models. The comparison shows that the prediction model has better performance in terms of applicability and accuracy

which can provide a reliable reference for specific heat capacity determination in theoretical calculations

numerical simulations

and engineering practice.