Nonlinear Model Predictive Control for Cost Optimal Startup of Steam Power Plants

摘要

The startup of processes is a challenging control task as a large range of operation needs to be passed and as multiple process units need to be coordinated. This makes the control problem nonlinear and multi-variable, respectively. Furthermore constraints on process variables have to be considered during a startup. Nonlinear Model Predictive Control (NMPC) is a promising concept to automate and optimize the startup of processes. Control actions are computed based on a nonlinear process model and employing a reasonable prediction horizon. This paper states the startup optimization problem and discusses the application of online optimization in an NMPC. Appropriate numerical solution methods and their implementation in a modern control system are motivated. The startup of a steam power plant serves as example. A process model is built using the object-oriented physical modeling technology Modelica. Based on an economic objective function the NMPC does both: online computation of optimal reference trajectories and generation of set points for an underlying base control system.