Y. Mollapour, M. Aghakhani, H. Eskandari, H. Azarioun2,
Volume 2, Issue 2 (Journal OF Welding Science and Technology of Iran 2016)
Abstract
This paper investigates the effect of boehmite nano-particles surface adsorbed byboric acid (BNBA) along with other input welding parameters such as welding current, arc voltage, welding speed, nozzle-to-plate distance on weld penetration. Weld penetration modeling was carried out using multi-layer perceptron artificial neural network (MPANN) technique. For the sake of training the network, 70% of the obtained data from experimentation using five-level five-factor central composite rotatable design of experiments was used. The performance of the network shows a good agreement between the experimental data and the data obtained from the network. Hence, it is to be concluded that MPANN is highly accurate in predicting the weld penetration in SAW process.
F. Pahnaneh , M. Aghakhani *, R. Eslami Farsani, M. Karamipour1,
Volume 6, Issue 1 (Journal OF Welding Science and Technology 2020)
Abstract
This paper reports the applicability of fuzzy logig (FL) to predict the hardness of melt zone (HMZ) during the gas metal arc welding (GMAW) process, which is affected by the combined effect of ZrO2 nano-particles and welding input parameters. The arc voltage, welding current, welding speed, stick-out, and ZrO2 nano-particles were used as the input parameters and HMZ as the response to develop FL model. The predicted results from FL were compared with the experimental data. The most important input parameter affecting the HMZs was the addition of ZrO2 nanoparticle coatings with a thickness of 1 mm, which increased the hardness from 78 to 84 HRB. The correlation factor value obtained was 99.98% between the measured and predicted values of HMZ. The results showed that FL is an accurate and reliable technique for predicting HMZ because of its low error rate. Also, the presence of ZrO2 nano-particles in the weld pool has increased the penetration up to 2 times.
