This article presents a study of the viscoelastic behavior of an epoxy polymer and a glass-reinforced composite based on it under cyclic thermomechanical loading. The goal is to model and explain the experimentally observed stress state formation, including the accumulation of residual stresses under various initial mechanical stress levels and heating/cooling cycle durations. An improved material model, implemented as a Python script, is used, allowing for the consideration of memory effects on thermomechanical loading depending on the level and nature (mechanical or thermal) of the initial stresses. A Python script was developed to determine the viscoelastic parameters (elastic modulus E1, elastic parameter E2, and viscosity) for the three-element Kelvin-Voigt model. These parameters were determined at different temperatures for both the polymer and the glass-reinforced composite used in the modeling. The accumulation of stresses under different ratios of mechanical and thermal stresses was also investigated. Experiments showed that high levels of residual stress could form in the pure epoxy polymer. The initial stress state significantly influences residual stress accumulation in the pure epoxy polymer. Low initial tensile stresses (0-1.5 MPa) resulted in substantial residual stress accumulation, exceeding the initial stresses by up to 2.7 times and reaching values of up to 2.1 MPa. Conversely, high initial stresses (around 4 MPa) suppressed residual stress accumulation due to the dominance of relaxation processes. This highlights the critical role of the initial loading conditions in predicting long-term material behavior. In the glass-reinforced plastic, the effect of residual stress accumulation was significantly weaker, possibly due to the reinforcement and high residual stiffness, even at elevated temperatures (the studies were conducted from 30 to 180 degrees C for the composite and from 30 to 90 degrees C for the polymer). The modeling results show satisfactory qualitative and quantitative agreement with the experimental data, offering a plausible explanation for the observed effects. The proposed approach and tools can be used to predict the stress-strain state of polymer composite structures operating under cyclic thermomechanical loads.
