IG-ENT:A innovative ensemble approach for the flow prediction of main steam system in thermal power plant

The main steam flow is a crucial indicator for monitoring and controlling industrial operations, directly affecting the accuracy of calculations for key metrics such as heat consumption rate and coal consumption rate. Traditionally, main steam flow monitoring relies on the Flügel formula for calculations; however, the predictive accuracy of this method is limited. Alternatively, installing traffic monitoring devices increases throttling losses. To address these challenges, we propose a data-driven integrated framework and establish an innovative ensemble model. In the data processing phase, we employ the mRMR method for feature selection. Furthermore, we use genetic algorithms to optimize the ensemble model. To enhance predictive performance, we propose ensemble models based on quantiles. Finally, we apply various approaches to validate the model's performance. Comparative results show that, compared to baseline models, integrated models, and baseline models integrated with quantile intervals, our model demonstrates superior predictive performance.