Modeling and optimization of dynamic-mechanical properties of hybrid polymer composites by multiple nonlinear neuro-regression method

The purpose of this research is to improve the dynamic-mechanical properties of the polypropylene filled by artichoke stem (AS) particles and wollastonite (W) in different weight fractions. The effect of weight ratios of fillers in polypropylene was mathematically modeled using the data obtained as a result of the experimental work. In the modeling phase, multiple nonlinear neuro-regression analysis was used. In this context, proposed linear and nonlinear models have been examined by performing R2training, R2adjusted, R2testing, and boundedness check. The models that satisfy these four criteria were selected as the objective functions for the optimization phase. Finally, Modified Differential Evolution Algorithm was used to obtain maximum storage modulus and loss modulus by adjusting weight percent ratio of artichoke stem particle and wollastonite. The experimental results and the modeling optimization results showed that when the polypropylene-artichoke stem particle-wollastonite hybrid polymer composite was used instead of other non-hybrid polymer composite, the storage modulus and the loss modulus improved by approximately 40%.