Based on the control of the whole production process, to make emissions of SO₂ concentration conform to the standards in the production of sulfuric acid with pyrite, we propose prediction for SO₂ concentration of tail gas with GA-ELM modeling. In the production of sulfuric acid by pyrite, we collect the key parameters with greater influence on the SO₂ concentration of exhaust gas, and use GA-ELM neural network to predict the SO₂ concentration while producing acid with exhaust gas. The method is tested in a factory, and the predicted results are highly identical with actual data. It plays an important guiding role for the adjustment and optimization of process index and exhaust emissions conforming to the standards.