Search

Highlights: [•] k-Step-ahead prediction error model identification is studied. [•] It will be shown that a model estimated using a k-step-ahead prediction error criterion may not be optimal for k-step-ahead prediction. [•] A normal one-step-ahead prediction error criterion can be optimal. [•] For MPC identification it is suggested that one-step-ahead prediction error criterion is used. [•] Simulations and industrial data will be used to illustrate the relevance of the result. [ABSTRACT FROM AUTHOR]

Abstract: The high cost of model predictive control (MPC) technology has hampered its wide application in process industries beyond the refining/petrochemical industry. This work strives to increase the efficiency of MPC deployment. First, a semi-automatic MPC system is introduced. It consists of three modules: an MPC module, an online identification module and a control monitor module. The goal of the MPC technology is twofold: (1) to considerably reduce the cost of MPC commissioning and maintenance and (2) to increase control performance. System identification plays important roles in all the three parts of the MPC system. In the identification module, the so-called ASYM method of identification is used. It is demonstrated with an industrial application. In the control module, adaptive disturbance model identification is developed for improving control performance; in the monitor module, a method of model error detection method is developed. Industrial applications and simulations are used to demonstrate the ideas. Finally, we comment on some industrial needs on MPC research and development. [Copyright &y& Elsevier]

Abstract: Identification of systems with slowly and irregularly sampled output data is considered. The identified models can be used in inferential control or as soft sensor. Dual-rate system with slow output samples is a special case of the problem studied here. In this work, it will be show that an output error method is a natural choice for the problem that can identify the fast rate model directly from the fast input and slow and irregular output data. When the system is in the model set, consistence of the output error model is established and minimum variance property is proved. When the model order is lower than the true order, bias distribution in the frequency domain is given. Simulation studies and industrial data will be used to illustrate the method. [Copyright &y& Elsevier]

Abstract: This work introduces a method of multivariable model error detection in model prediction control (MPC). The idea is to use non-disturbing small sinusoidal test signals to obtain accurate estimates of process frequency responses at several frequency points. Then, the differences between estimated frequency responses and the frequency responses of current MPC model are used to form the model error index matrix which is used to access the model error of the MPC controller. An upper error bound is developed for quantifying the error of frequency response estimation. The method works in closed-loop operation with the MPC controller online. Simulation studies are used to demonstrate the use of the method. [Copyright &y& Elsevier]