Experimental, Modeling, and Optimization Investigation on Mechanical Properties and the Crashworthiness of Thin-Walled Frusta of Silica/Epoxy Nano-composites: Fuzzy Neural Network, Particle Swarm Optimization/Multivariate Nonlinear Regression, and Gene Expression Programming

In this work, an experimental study on the quasi-static collapse of thin-walled frusta of silica/epoxy nano-composites was conducted. The effect of nano-silica content and the particle size hybrid on the energy absorption capability of thin-walled frusta, the impact strength, Young's modulus, and the yield strength was investigated. For this purpose, three various sizes of the silica particle with the mean diameter of 17, 25, and 65 nm were used. The results showed that by adding the silica nano-particles up to 6 wt.%, the impact strength and Young's modulus increased, the yield strength remained constant, and the crashworthy capability of structures decreased. Also, two approaches including Fuzzy Neural Network, the hybrid of Particle Swarm Optimization (PSO), and Multivariate Nonlinear Regression (MNLR) were employed to determine the effect of the mentioned parameters. In comparison with the mentioned models and the experimental results, PSO/MNLR approach showed a better prediction for the parameters. Different parameters were optimized by Gene Expression Programming. Some fracture surfaces were studied by scanning the electron microscopy.

Automated summary

The experimental study investigated the quasi-static collapse of thin-walled frusta of silica/epoxy nano-composites, finding that adding silica nano-particles increased impact strength and Young's modulus but decreased crashworthiness. The PSO/MNLR approach had better prediction accuracy for parameters compared to other models. Fracture surfaces were analyzed using scanning electron microscopy.