A predictive control strategy for nonlinear NOx decomposition process in thermal power plants

For the load-dependent nonlinear properties of the nitrogen oxide (NOx) decomposition process in thermal power plants, a local-linearization modeling approach based on a kind of global Nonlinear AutoRegressive Moving Average with eXogeneous input (NARMAX) model, named the exponential ARMAX (ExpARMAX) model, is presented. The ExpARMAX model has exponential-type coefficients that depend on the load of power plants and are estimated offline. In order to take advantage of existing conventional controllers and to reduce the cost of the industrial identification experiment, we propose a model structure that makes it possible for the ExpARMAX model to be identified using commercial operation data. On the basis of the model, a long-range predictive control strategy, without resorting to parameter estimation online, is investigated. The influence of some intermediate variables treated as process disturbances is studied, and the scheme using a set of multi-step-ahead predictors of the intermediate variables to improve control performance is also presented. A simulation study shows that the ExpARMAX model can give satisfactory modeling accuracy for the NOx decomposition (de-NOx) process in a large operating range, and the control algorithm proposed significantly improves the control performance