摘要：AbstractAt Lonmin’s base metal refinery a blending strategy is implemented to minimise variations in the mass flow of silica, iron, selenium, and tellurium into the plant. These impurities contaminate the plant’s products and control adjustments are made throughout the plant to minimise their impact. Blending smoothes the dynamics of these impurities at the input of the plant, which aids plant-wide control and ultimately improves the plant’s revenue. This work presents an (µ + λ)-evolutionary strategy to optimise the plant’s input blend and assesses its performance using historical data. With the evolutionary strategy it was possible to achieve an 84% improvement over an unblended feed and a further 48% improvement over a randomised feed strategy.
