Electrical Properties Research of PA6 Low Warp Composite Material Using Selected Machine Learning Method

Abstract

This paper focuses on the study of the electrical properties of PA6 Low Warp material using Artificial Neural Network model. The model classifies the material samples into a given class depending on the time of exposure of the material to increased temperature according to the input electrical properties and the mechanical properties of the material in tension. For the neural network models tested, the primary effectiveness of the ReLu and Tanh activation functions to classify a sample into a given class for a given time interval at an increased environment temperature of 180 °C was examined. The highest classification accuracy of 82.46% was obtained for the model using the Tanh activation function. The results show that in researching the physical properties of engineering materials used in 3D printing using artificial neural networks allows to predict the response of material properties under certain initial ambient conditions.