Aiming to address the issue that the performance of the offline model for an ultra‐supercritical coal‐fired power plant boiler gradually degrades after being used for a period under varying operating conditions, this article proposes an online prediction model for NO x concentration based on an adaptive online support vector regression (Online‐SVR) algorithm. Using actual unit measurement data, an online modeling approach is developed. On the basis of the traditional incremental SVR learning algorithm, training samples are selectively added and removed according to the similarity between samples. By continuously learning from the newly added samples, the model enhances its ability to track the change of process characteristics on the foundation of the pre‐existing offline model. The results demonstrate that the proposed model significantly improves the prediction accuracy and computational speed compared to offline models. It effectively enhances the model’s ability to learn from new samples and meets the timeliness requirements for online operation in coal‐fired power plants.
Zhang et al. (Wed,) studied this question.