Modeling and simulation of controlling the mechanical properties of polymer composites using heuristic and meta-heuristic algorithms

The strength of a material is equal to the maximum stress that that material can withstand under uniform tension, which is sometimes equal to the yield stress and sometimes equal to the tearing stress of that material. In the case of polymer composites that include mineral particles with micro or nano size, the strength of the compound is affected by the stress transfer between the matrix and the fillers, which is affected by factors such as particle size, amount of particles, and the compatibility of the particles with the polymer matrix. An alternative model based on An artificial neural network has been developed to predict the stress reduction of polymer matrix composite. The important point in choosing polymer matrices in polymer matrix composites is their sensitivity to moisture. Resins tend to absorb water, and this causes dimensional changes and a decrease in strength and stiffness at high temperatures. The amount of moisture absorption, which is usually measured as a percentage of weight gain, depends on the polymer and the relative humidity. When resins are placed in a dry environment, they repel moisture. The amount of adsorption and desorption strongly depends on the temperature. The moisture sensitivity of resins is very different, so that some of them are very resistant. The ANN model is trained and validated with ۹۰۰۰ experimental data sets obtained from stress relaxation tests under different conditions of constant strain (initial stress) and constant temperature. ANN training uses a scaled conjugate gradient method. Brain surgeon's optimal algorithm is used for topology optimization. The optimal ANN configuration has ۸۸ processing elements (۳ in the input layer, ۴۵ in the first hidden layer, ۳۹ in the second hidden layer, and ۱ in the output layer) and ۴۱۰ links. ANN model predictions are more accurate in a wider range of stress and temperature.

کلیدواژه ها:

Nonlinear viscoelastic ، intermediate phase ، micromechanical modelingNeural network ، Polymeric composite