Optimization of chemical composition in the manufacturing process of flotation balls based on intelligent soft sensing

Abstract

This paper presents an application of computational intelligence in modeling and optimization of parameters of two related production processes - ore flotation and production of balls for ore flotation. It is proposed that desired chemical composition of flotation balls (Mn=0.69%; Cr=2.247%; C=3.79%; Si=0.5%), which ensures minimum wear rate (0.47 g/kg) during copper milling is determined by combining artificial neural network (ANN) and genetic algorithm (GA). Based on the results provided by neuro-genetic combination, a second neural network was derived as an ‘intelligent soft sensor’ in the process of white cast iron production. The proposed ANN 12-16-12-4 model demonstrated favourable prediction capacity, and can be recommended as a ‘intelligent soft sensor’ in the alloying process intended for obtaining favourable chemical composition of white cast iron for production of flotation balls. In the development of intelligent soft sensor data from the two real production processes was used. [Projekat Ministarstva nauke Republike Srbije, br. TR35037 i br. TR35015]

Keywords

• wear rate
• chemical composition
• neural networks
• genetic algorithm
• optimization