Chemical processes are always of great dynamic fluctuations and uncertainty.An on-line feedback optimization with model uncertainty was presented to resolve the problem.The process dynamic model was discretized as difference equations according to a specific period

and dynamic optimization was carried out based on the difference equations.The profit at the end of optimization horizon was used as optimization objective

manipulation variables were used as optimization variables

a non-linear program was used as optimization algorithm.On-line feedback was implemented according to measured variables after the optimization results were adopted.In the difference equations correction terms were introduced

which could be calculated when the actual values of process variables were imported into the difference equations.Simulations showed that compared with steady optimization

feedback optimization not only could also make the process run at the optimum operating point but also could have high real-time quality.