Faculty
Department
Year of Publication
upload
Publication Type
Abstract
The integrity of welded structures is affected by weld defects, induced stress as well as its
resistance to varying impacts during and after fabrication. This study explores the application of
the Adaptive Neuro-Fuzzy Inference System (ANFIS) in optimizing and predicting the impact
toughness of Tungsten Inert Gas (TIG) welded mild steel joints aimed at enhancing weldment
quality and overall structural integrity by determining the influence of key welding process
parameters on the impact toughness of the resultant weldment. The research seeks to optimize
predict these relevant factors thereby addressing challenges such as induced stress and failures
resulting from impacts on weldment.
Central Composite Design (CCD) was employed for experimental design having current, voltage
and gas flow rate as weld process generating twenty (20) experimental runs. Mild steel plates were
cut and welded using a TIG welding equipment to produce weld samples using the varying process
parameters. A digital impact testing machine was used to measure the impact toughness of the
weldments. The experimental data was then analyzed using ANFIS, which integrates neural
networks and fuzzy logic for predicting and optimizing the investigated response.
The ANFIS model effectively trained and tested the experimental data after which an optimal
result having having current of 175 amps, voltage of 23.5 volts, and a gas flow rate of 15.5
liters/min would yield a maximal impact toughness values of 95.7 J. Post experimental results
shows high correlation values with the optimal result thereby serving as validation. These findings
underline the potential of ANFIS as a robust tool for advancing production engineering processes.
This result improves the reliability of welded structures and supports the advancement of
production engineering practices
resistance to varying impacts during and after fabrication. This study explores the application of
the Adaptive Neuro-Fuzzy Inference System (ANFIS) in optimizing and predicting the impact
toughness of Tungsten Inert Gas (TIG) welded mild steel joints aimed at enhancing weldment
quality and overall structural integrity by determining the influence of key welding process
parameters on the impact toughness of the resultant weldment. The research seeks to optimize
predict these relevant factors thereby addressing challenges such as induced stress and failures
resulting from impacts on weldment.
Central Composite Design (CCD) was employed for experimental design having current, voltage
and gas flow rate as weld process generating twenty (20) experimental runs. Mild steel plates were
cut and welded using a TIG welding equipment to produce weld samples using the varying process
parameters. A digital impact testing machine was used to measure the impact toughness of the
weldments. The experimental data was then analyzed using ANFIS, which integrates neural
networks and fuzzy logic for predicting and optimizing the investigated response.
The ANFIS model effectively trained and tested the experimental data after which an optimal
result having having current of 175 amps, voltage of 23.5 volts, and a gas flow rate of 15.5
liters/min would yield a maximal impact toughness values of 95.7 J. Post experimental results
shows high correlation values with the optimal result thereby serving as validation. These findings
underline the potential of ANFIS as a robust tool for advancing production engineering processes.
This result improves the reliability of welded structures and supports the advancement of
production engineering practices
Supervisor(s)
co-supervisor