Otimização paramétrica da haste de um drone utilizando algoritmo genético

Abstract

The use of drones involves many areas and applications, from monitoring agricultural spraying to shipment of goods. For this, getting the best overall performance is essential, and this is achieved through optimization projects, especially in the rods (the largest part of the struc tural net volume). Therefore, this work performs a parametric optimization (PO) of the arm of a drone, subjected to the efforts of a selected rotor-propeller assembly, aiming to achieve the multiobjective of minimizing mass and maximizing stiffness, using a genetic algorithm (GA) as an optimizing method. Initially, the complex geometry of the standard model is sim plified and a Python script is developed that generates a rod model that can be analyzed using the finite element method (FEM). It is developed in such a way as to identify nine design variables and each change in their values leads to the generation of a new finite ele ment (FE) models: width L; height Ha; thicknesses of the truss sections Tt , of the edges Te, of the lower Trl and upper Tru stiffener; number of repetitions of the “cell” of the truss Nrep; and widths of the lower Rl and upper Ru stiffeners. Then, a sensitivity study is carried out to find out which of the selected variables have the greatest influence on the mechanical res ponse of the rod subjected to the loads of the rotor-propeller assembly. Based on this study, some of the less influential variables are eliminated in order to proceed with development of the script to establish communication between the EF model and the optimization algorithm. After defining the population and generation size, the PO is run to obtain the best values for the design variables inside the upper and lower limits, and then select the best ones and analyze them against each other. Finally, after checking the simplified model from the sensitivity study, the Tru and Rl variables were eliminated because they had little influence on the mechanical response. After completing the OP, some models were selected, with the highest mass minimization of 28.36% and the highest stiffness maximization of 277% (to the detriment of a slightly reduced mass) in different models. There is also another with a balance between functions, obtaining a 15.31% reduction in mass and a 156% increase in stiffness, and finally, one that had its CS reduced in relation to the standard, decreasing from 33 to 30, but still being far from failure.