Multi-Output Soft Sensor with a Multivariate Filter That Predicts Errors Applied to an Industrial Reactive Distillation Process

The paper deals with the problem of developing a multi-output soft sensor for industrial reactive distillation process methyl tert-butyl ether production. Unlike existing approaches, this proposes using filters to predict model errors, which are then taken into account as corrections in final predictions outputs. decomposition optimal estimation time delays is proposed each input sensor. Using approach concentrations sec-butyl ether, methanol, and sum dimers trimers isobutylene output product column was shown improve results by 32%, 67%, 9.5%, respectively.