MACHINE LEARNING-ASSISTED PREDICTION AND VIKOR-BASED OPTIMIZATION OF MECHANICAL PROPERTIES IN MICRO-FRICTION STIR WELDED MONEL 400 SHEETS
Srujan Manohar MVN, Mahadevan K, Arun P, Varsala Pragna, Mallikarjunachari G
Abstract: Micro-friction stir welding (μ-FSW) of 0.6 mm Monel 400 sheets was investigated for butt and lap configurations using a tungsten-carbide tool. Four experimentally produced weld profiles were evaluated for tensile properties, micro-hardness, surface roughness and weld-nugget microstructure. Machine-learning models were used as exploratory tools to study relationships between process variables and measured responses; experimental observations and synthetic augmented records are explicitly distinguished in the revised methodology. Synthetic records are not treated as independent experiments and are not used as the basis for the VIKOR decision matrix. The experimental decision matrix was analysed using the VIKOR compromise-ranking procedure with equal criterion weights. Using ultimate tensile strength, yield strength, elongation and hardness as benefit criteria and surface roughness as a cost criterion, profile L2 ranked first (Q = 0.000), followed by L1, B2 and B1. A ±20% one-at-a-time weight perturbation around equal weighting retained the same ranking. The study therefore identifies L2 (800 rpm, 20 mm/min, 5.5 kN) as the best compromise among the four tested conditions while clearly limiting conclusions to the available experimental dataset.
Keywords: Micro-friction stir welding, Monel 400, machine learning, VIKOR, tensile properties, micro-hardness, surface roughness
DOI: 10.24874/PES08.03B.014
Recieved: 11.03.2026 Revised: 01.06.2026 Accepted: 10.07.2026
UDC:
Reads: 8 
