Enhanced investigations and modeling of surface roughness of epoxy/Alfa fiber biocomposites using optimized neural network architecture with genetic algorithms

Currently, there is a notable attraction within the industry towards biocomposites, driven by the increasing fascination with natural fiber-reinforced composites (NFRCs). These NFRCs offer remarkable benefits, including cost-effectiveness, biodegradability, eco-friendliness, and favorable mechanical properties. As a result, the manufacturing processes of natural fiber reinforced polymer (NFRP) composites have garnered attention from both industrial professionals and scientists. The emergence of these eco-friendly materials in the automotive and aerospace industries has sparked interest in understanding their production techniques. However, the machining processes of NFRP composites pose significant challenges due to the complex structure of natural fibers, necessitating thorough studies to address these issues effectively. This research paper presents a comprehensive investigation on surface roughness during the milling process of Alfa/epoxy biocomposites. A set of 100 experimental trials was conducted to test the surface roughness, and analysis of variance (ANOVA) was used to assess the impact of cutting parameters and chemical treatment on surface quality.To develop a predictive model for surface roughness, a hybrid approach called ANN-GA (artificial neural networks-genetic algorithms) is proposed in this research. This approach combines ANN and GA to determine an optimal neural network architecture. The performance of the ANN-GA model is compared to the Levenberg-Marquardt backpropagation (LM) algorithm.ANOVA results show that the feed per revolution have a significant influence on surface roughness, followed by the chemical treatment of fibers, while machining direction has a smaller effect. The ANN-GA model demonstrates good accuracy in surface roughness prediction compared to the LM algorithm.