Estimation of hanging and removal times in eloxal with artificial neural networks

In an enterprise that produces aluminum (Al) profiles, the hanging and removal processing times of profiles coming to the eloxal department vary. Due to the variability of the profile time, more or fewer orders are received from customers and problems with delivery times are experienced. In this study, the hanging and removal processing times of any profiles coming to the eloxal department of the company producing aluminum profiles were estimated with artificial neural networks (ANNs). Since there were no recorded transaction times in the company, firstly the hanging and removal processing times were measured with a time study. Regression analysis determined that the most important factors affecting the hanging and dismantling process times were profile circumference, d _u and square meters. These three factors were used as inputs, and an ANN model was established. While the hanging and removal processing times in the eloxal department were estimated by the ANN, 70% of the data were used for training, 20% for verification and 10% for testing. The hanging and removal processing times obtained by the ANN model were compared with the results obtained by the time study. The ANN model, which was developed according to statistical results, can accurately predict hanging times by 99% and dismantling times by 96%.