Abstract
A number of evolutionary algorithms such as genetic algorithms, simulated annealing, particle swarm optimization, etc., have been used by researchers in order to optimize different manufacturing processes. In many cases these algorithms are either incapable of reaching global minimum or the time and computational effort (function evaluations) required makes the application of these algorithms impractical. However, if the Nelder Mead optimization method is applied to approximate solutions cheaply obtained from these algorithms, the solution can be further refined to obtain near global minimum of a given error function within only a few additional function evaluations. The initial solutions (vertices) required for the application of Nelder-Mead optimization can be obtained through multiple evolutionary algorithms. The results obtained using this hybrid method are better than that obtained from individual algorithms and also show a significant reduction in the computation effort.
Highlights
In recent years, optimization of various manufacturing processes using machine learning and evolutionary algorithms are becoming a key requirement for various industries
This paper demonstrates a reduction in this effort and time by initially finding approximate solutions using any of the above mentioned algorithms and applying the Nelder-Mead optimization (NMO) method to further refine them
From the experimental data obtained and the computational models developed, the following can be concluded: 1. A number of evolutionary algorithms can be used for optimization of weld bead geometry of a tungsten inert gas (TIG) welding process using a filler material
Summary
Optimization of various manufacturing processes using machine learning and evolutionary algorithms are becoming a key requirement for various industries. Et al [14] applied various optimization algorithms (GA, SA and PSO) in order to optimize friction-stir welding process parameters to obtain desired tensile strength and minimize metal loss. They found that among the algorithms they used, GA outperformed all the other algorithms and the results obtained from GA had a good agreement with the experimental data. Et al [15] compared GA and RSM in order to optimize the welding process based on four quality responses (deposition efficiency, bead width, depth of penetration and reinforcement) They found that GA can perform better than RSM; optimization using GA requires a good setting of its internal parameters. NMO is an optimization method that is most effective when it is unconstrained, and cannot be applied directly to the optimization problem in many of the cases as it can lead to an infeasible solution
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have