A back forward Artificial Neural Network (ANN) was used to study the effect of flotation time, collector dosage, frother dosage, and impeller speed on flotation recovery and grade. The results of 13 flotation experiments conducted on Egyptian coal were used for training the network while another 28 experiments were used for validation. Simulation results showed that a one layer network with a [6 2] architecture was the one that gave the least standard error (SE). The values of this error were 6.66% and 0.37% with recovery and grade, respectively. The software used in this paper was designed to automatically select the network architecture according to the direction and position of the network error. Using this ANN to optimize the flotation process showed that the optimum flotation parameters were 207.6 seconds for the flotation time, 1865.6 g/t for the collector dosage, 828 g/t for the frother dosage, and 1754 rpm for the impeller speed. The results of optimization process showed that experiment number 135 gave the highest recovery and grade: 94.94% and 5.07% respectively.
Real Time Impact Factor:
1
Author Name: M. Farghaly, A. Serwa, M. Ahmed
URL: View PDF
Keywords: Neural Network; Froth Flotation; Optimization; Coal Beneficiation
ISSN: 2169-642X
EISSN: 2169-6438
EOI/DOI:
Add Citation
Views: 1013
