Structural safety is directly affected by the mechanical properties of concrete at high temperatures. Firstly, based on the existing compression and tension test data of concrete at high temperatures, the Abaqus finite element software was adopted for numerical simulation reproduction, and the reliability of the simulation method was verified. Secondly, by simulating the uniaxial tension-compression and confining pressure tests of normal concrete with different strength grades under high temperatures of 20?800 ℃, the influence rules of temperature on the compressive strength, splitting tensile strength, elastic modulus, and stress?strain relationship of concrete were elucidated. Finally, based on three commonly used machine learning algorithms, i.e., BP neural network (BPNN), support vector regression (SVR), and Gaussian process regression (GPR), a rapid prediction model for the mechanical properties of concrete at different temperatures was established. The results indicate that the prediction accuracies of the GPR and SVR models are relatively high, and the R² of the prediction models for the compressive strength and tensile strength of concrete are 0.997 23 and 0.979 55, respectively.