Abstract

A difficult task for the transport sector is to make its assemblies lighter and perform more efficiently. Use of aluminum and its alloys has increased extensively in this sector because of reduction in weight of the vehicles and resulting energy savings. High thermal conductivity and thermal expansion pose difficulty in welding of these alloys. Cold metal transfer (CMT), a low heat input welding process, is the best choice for welding of these alloys. However, controlling the welding input parameters is highly necessary to obtain defect-free and high strength welded joints. In the present study, an attempt is made to develop a Matlab software-based application by two approaches, such as multiple regression analysis (MRA) and adaptive neuro-fuzzy inference system (ANFIS), for predicting the complete weld bead shape (graphical representation) of AA5052 using the CMT welding process. The data inputs used for both approaches are welding current (A) and welding speed (mm/min), respectively. A graphical interface is built to help the user to choose welding input parameters and obtain directly a representation of the weld bead profile in graphical form. In addition, the output response shows the complete weld bead shape, which is defined by the X and Y coordinates of the various points in the weld bead profile. The results are validated with randomized tests against the weld bead shape predicted by Matlab. Comparatively, ANFIS is the more effective method for predicting the weld bead profile and shows better agreement with the experimental profile than MRA. Further, the reliability and stability of the ANFIS model were determined from the mean absolute error percentage, root mean square error values, and linear R2fit model, confirming that the ANFIS-based prediction is in better agreement with the experimental values than MRA.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call