A genetic algorithm optimization framework for the characterization of hyper-viscoelastic materials: application to human articular cartilage

Piers Allen,Sophie C. Cox,Simon Jones,Daniel M. Espino
DOI: https://doi.org/10.1098/rsos.240383
IF: 3.5
2024-06-26
Royal Society Open Science
Abstract:This study aims to develop an automated framework for the characterization of materials which are both hyper-elastic and viscoelastic. This has been evaluated using human articular cartilage (AC). AC (26 tissue samples from 5 femoral heads) underwent dynamic mechanical analysis with a frequency sweep from 1 to 90 Hz. The conversion from a frequency- to time-domain hyper-viscoelastic material model was approximated using a modular framework design where finite element analysis was automated, and a genetic algorithm and interior point technique were employed to solve and optimize the material approximations. Three orders of approximation for the Prony series were evaluated at N = 1, 3 and 5 for 20 and 50 iterations of a genetic cycle. This was repeated for 30 simulations of six combinations of the above all with randomly generated initialization points. There was a difference between N = 1 and N = 3/5 of approximately ~5% in terms of the error estimated. During unloading the opposite was seen with a 10% error difference between N = 5 and 1. A reduction of ~1% parameter error was found when the number of generations increased from 20 to 50. In conclusion, the framework has proved effective in characterizing human AC.
multidisciplinary sciences
What problem does this paper attempt to address?