The Feed Forward Neural Network Model for Liquid-Liquid Extraction and Separation of Cobalt (II) from Sodium Acetate media using Cyanex 272

摘要

Cobalt is one of the precious ferromagnetic metals, which widely used in the preparation of magnetic, wear-resistant and high-strength alloys. This metal was not found naturally in single metal form but is found as impurities in nickel or copper ore. The extraction process is one of the methods to separate cobalt from its impurities. However, this process needs an expensive organic solution. In practice, changing the composition of chemicals composition in extraction process always affect at a high cost. Therefore, the development of the artificial neural network (ANN) model to model the cobalt extraction process can serve as an important tool for predicting and investigating the optimum production for the cobalt extraction without the need to run the actual experiment. Hence, the development of the ANN model of cobalt extraction model is essential to simulate the process, which can lead to high yields of cobalt production. In this work a selected optimum multiple-input-single-output (MISO) model of feed forward neural network (FFNN) was used to predict the percentage of cobalt extraction. MISO FFNN with 20, 30 and 50 hidden nodes were used to simulate cobalt extraction process. The simulation results achieved was compared with data available in the literature. The results show that MISO FFNN with 50 hidden nodes has the best performance. The optimum result of MISO FFNN then exported to Simulink model in Matlab environment, hence make it easy to use in predicting and investigating for the optimum production of the cobalt extraction.