Optimization Of Robotic P-Gtaw Welding Process Parameters For Welding Ni-Cr Steel Alloy By Combining A Deep Neural Network And Multi-objective Genetic Algorithm

NSGA-type algorithms are widely used in mathematical optimization to power up diverse strategies for hyper-heuristics methodologies in problem-solving, combinations of these genetic algorithms and other techniques are also well documented in the literature related to the use of alternative ways to solve situations in the leading industries. In critical welding processes like armouring, it is essential to achieve excellent solutions due to the nature of the usage of the piece, whose principal objective is the protection and preservation of life. In armouring applications, the weakest point of the entire piece is the soldering strip itself.

In this particular application, using a robotic-guided welding process, the idea is to combine the NSGA-II with neuronal networks to find out the most impacting and optimal welding variables values to preserve the best mechanical properties of the resulting prototypes, by the use of the design of experiments (DOE), neuronal networks and multiobjective genetic algorithms the aim of this work is to present the pulsed Gas Tungsten Arc Welding (P-GTAW) process optimization by using a new approach combining the aforementioned to find the optimal values of three process key variables.

After intensive experimentation using different techniques of parameter estimations (traditional for this process, and the by the utilisation of the new approach), the results of the analysis are presented and compared, visual inspection and other inspections point to "attractive solutions" for welding with the new methodology