Simulation of Thermoset Heat Conductivity by Means of Artificial Neural Networks

The study is relevant due to necessity of improvement of thermal and physical features of composite materials for ensuring the size stability of composite structures. Therefore, this Article is aimed at developing an automatized system for determining the thermal conductivity of thermosetting polymers by modified nanosized carbon structures depending on the process parameters. Since this is a multiple-parameter task lacking formalized solution methods which could ensure a high performance capability and fair accuracy, the method of simulation by means of artificial neural networks was used. This Article describes the procedure of manufacture and research of thermal conductivity of specimens of nano-modified binder. Upon the experimental research results, databases of thermal and physical features and manufacture process parameters of thermosetting polymers were formed what enabled the formation of a neural network model of dependence of thermal and physical features of thermosetting polymers modified with nanosized carbon structures from the manufacture process parameters. The elaborated neural model of dependence of thermal and physical features of thermosetting polymers modified with nanosized carbon structures ensures the total convergence of the obtained values for samples of the tuition, validation and testing -98.77%.