An ANN Approach in Prediction of Microhardness in a Furnace Cooled Sintered P/M 6061 Aluminium Compacts

Abstract

In this paper the effect of particle size of aluminium powder and furnace controlled cooling after sintering on porosity level and micro hardness of an elemental 6061 aluminium alloy has been investigated experimentally and the micro hardness value is compared with the Neural network algorithm using matlab. The algorithm used here are Gradient Descent Back propagation with Adaptive Learning Rate. Aluminium particle sizes of 20µm and 150µm were used. The elemental 6061 aluminium powders are warm compacted at 175MPa. After sintering for about one hour at 600°C, the aluminium compacts were furnace cooled at the rate of 1°C/min to different temperatures of 500°C, 400°C, 300°C and 200°C. When the cooling temperature after sintering inside the furnace is effected at various temperatures from 600°C to 200°C, for a precipitate hardened aluminium compacts with aluminium particle size of 20µm, the porosity level reduced by 26% and that for aluminium particle size of 150µm, the porosity level reduced by 23%. Marked improvement in micro hardness value is also observed correspondingly. Then the Neural Network was trained using the prepared training set which was recorded by the experimental values. At the end of the training process, the test data were used to check the accuracy result. As a result the Neural Network was found successful improvement in prediction of microhardness in a slow cooled sintered powder metallurgical 6061 Aluminium alloy.