PROGRAMAÇÃO MULTIOBJETIVO APLICADA AO DESENVOLVIMENTO DE MATERIAIS AVANÇADOS

Abstract

In this work we discuss a multiobjective optimization problem to study the properties of titanium alloys that will be represented by mathematical models. The titanium alloy is coated by specific processes so that its biofunctional properties are improved. We propose two polynomial mathematical models to represent the hardness and bond strength of the alloy as a function of the input parameters of the coating process. These models are predictive, have been developed roughly based on experimental data provided by Rafieerad et al. (2017) and had minimization error minimized by the least squares method and linear programming approach. We optimize these models by implementing a heuristic algorithm to obtain values for the properties of the titanium in question. We present in this work a methodology to construct the models and to optimize them, through the methodology it is possible to compare, discuss and present ways of adjusting coefficients of models besides providing a heuristic algorithm for multiobjective computational optimization of these same models. The methodology can reduce the high costs with laboratory experiments, since only the most suitable computational results for adhesion strength and hardness would be tested and confirmed in the laboratory, avoiding tests of new uncertain configurations, input parameter configurations of the process that covered the titanium alloy. Regarding computational implementation, we follow software engineering principles to make the implementation extensible and malleable, allowing the addition of new models and testing their effectiveness. The numerical data obtained from the execution of the implementation are displayed by means of graphics through the integration with Wolfram Mathematica software. The best models proposed are those that will simultaneously provide the highest values for adhesion strength and hardness, since the multiobjective problem is of maximization. The proposed methodology is a positive response to the inefficient advanced material development scenario described in Rafieerad et al. (2017). New experimental data are used to readjust the coefficients of the models in an iterative way