An Automatic Load Increase Prediction Model for Supercritical Units Based on xLSTM-FCNN

This paper addresses the challenge of automatic load increase during startup—from grid connection to dry-state operation—of supercritical units. An xLSTM-FCNN-based load change rate prediction model is proposed, and a decision mechanism for coal feed rate adjustments is established to enable automatic load increase. First, historical operational data were analyzed and preprocessed, and Spearman correlation analysis was employed to select input and output variables for the model. Next, an xLSTM-FCNN model was developed to forecast the load change rate and generate coal feed rate adjustment commands. Finally, simulation experiments were conducted using operational data from a 350 MW supercritical unit. Experimental results demonstrate that, compared to conventional time series prediction models, the proposed method exhibits superior predictive performance and accurately outputs coal feed rate adjustment commands aligned with control targets, demonstrating practical application value for automatic load increase in supercritical units.