Use of Modern Means of Artificial Intelligence to Predict the Electrical Resistance of a Polymer Composite Filled with Anisometric Carbon Nanoparticles

The aim of the work was to construct a digital model of a composite material of the dielectric-conductor type and to use it to predict the specific volume resistance as the main output parameter of the studied material. The composite material was produced from a fiber-forming polypropylene matrix with anisomeric carbon nanoparticles, i.e., multi-walled carbon nanotubes, as the filler. The input parameters were established from the physical properties of the components and experimental data. A neural network model was constructed to predict the conductivity of the material. A scattering diagram was used to show that the trained neural network gave highly accurate predictions. A decision tree for finding the significance and combinations of values of input parameters corresponding to various levels of specific resistance was built.