A Nonlinear Gaussian Mixture Model for the Prediction of Condenser Vacuum in a Coal-Fired Power Plant

This paper presents a novel model for predicting the condenser vacuum in thermal power plants, leveraging Nonlinear Gaussian Mixture Regression (NGMR) to address the complex, multimodal characteristics of the system. First, mutual information is applied to identify the correlation between auxiliary variables and condenser vacuum, thus selecting relevant features for the model. With the identified variables, the NGMR model is constructed to predict condenser vacuum levels. The performance of the NGMR model is evaluated and compared with other models, demonstrating superior accuracy and generalization capability.