Prediction of Mechanical Properties of Synthetic Waste Reinforced Polyolefins with GA-LSTM Hybrid Model

Abstract

In this study, the effects of the production parameters used in injection molding of particle-reinforced thermoplastics on the product quality and mechanical properties of the produced part are modeled using an optimized Genetic Algorithm-Long Short Term Memory (GA-LSTM) hybrid deep learning method. Here, LDPE, HDPE, and PP, the most important members of the polyolefins group, were used as thermoplastics, while powdered synthetic paint wastes were evaluated as reinforcement elements. Using different parameters, 819 specimens were produced by injection molding, and mechanical tensile, three-point bending, and izod impact tests were performed on each specimen. The GA-LSTM model was trained with the parameters used and the results obtained during the experimental process, and the predicted values were determined to correspond to the actual values. Well-known methods were used to measure the success of the hybrid GA-LSTM model. The designed GA-LSTM model produced the best outcomes, according to the results attained.