Nonlinear Model Predictive Control for Fossil Power Plants

This paper details the development of practical, multivariate, nonlinear, model predictive control (NMPC) for fossil power plants. The approach used here involves the development of a first-principles, nonlinear, reduced-order model (ROM) which captures the dominant static and dynamic characteristics of a power plant. The parameters of the model are estimated using prediction error methods or nonlinear least squares. This model is then used in a Kalman filter to estimate process states in real time. These estimated states are used for prediction, enabling the computation of the optimal control sequence. Compared to linear forms of MPC, the nonlinear first-principles approach has the significant advantages of a greater operating range, and the capability to generate diagnostics for operators. The results of the full-scale boiler control simulation conclusively demonstrate feasibility of the approach, and appear to be significantly better than those of most existing control systems. Implementation and on-line testing at an operating plant is planned. The focus of this paper is the development of the ROM, since this is the key to successful implementation of the approach. A previous paper described the estimation/control methodology and results in more detail.