The paper presents a methodology for building a digital model of the thermo-deformation system of a machine tool used in the creation of digital twins of machine tools for which their thermoelastic behavior is relevant, based on the solution of the thermoelasticity problem. Ansys Workbench and Ansys Twin Builder platforms are used as special software that allows you to create a digital model. A double-sided face grinding machine was used as an object of modeling, for which experimental studies were conducted ealier. The experimental data allowed us to form the target values of the optimization problem. A general methodology for building a digital model based on the use of the FMI standard for the exchange of dynamic libraries has been developed. The paper shows the practical application of the digital model of the machine tool to solve the problem of identification of the thermo-deformation model of the machine tool. The identification problem was solved in an extremal formulation. The analysis of the effectiveness of the use of six optimization methods is presented: the Quasi-Newton method; Pattern search; Multi-Objective Genetic Algorithm; Mixed-Integer Sequental Quadratic Programming; Adaptive Multiple-Objective; Adaptive Single-Objective. The methods of Mixed-Integer Sequental Quadratic Programming and Adaptive single-Objective were preferred in terms of efficiency. Their application allows to obtain the refined parameters of the model with the error of achieving the target values of modeling less than 2 %.