Operation parameters optimization of butadiene extraction distillation based on neural network

In this paper, based on the material balance, we used Aspen plus software to make sensitivity analysis of separation performance of the key component and the operational parameters. We quantitatively analyze the influence between the component parameters and the process. At last, we used the neural network to forecast the solvent ratio on diffident C4 feed and calculated the optimization solvent. At the end of the paper, we used the actual production data to test the validity of the model. On the view of reducing the energy consumption and ensuring the product quality, the research result told us that solvent ratio could be reduced nearly one percentage point.