Optimized Operation of Nonlinear Processes - A Multi-model Approach

The development of dynamic models based on physics is one of the biggest problems that real time optimization (RTO) faces for industrial processes. In this work, we develop a new approach by using system identification for the closed-loop dynamics in the optimization algorithm to overcome such a problem. The identification is done by using transfer functions and Wiener models with piece-wise linear output functions to generate multi-model for different cases of disturbances or other measured external inputs. These models are used to calculate optimal references to be used in RTO. The new procedure is applied to a continuous stirred tank reactor (CSTR). The numerical simulations show the effectiveness of the implemented method by comparing it with previous work from literature.