. The construction of a novel electric power system imposes new requirements on the operational flexibility of thermal power units

necessitating frequent and significant load adjustments to become the new normal during operation. In the process of varying load in these units

the real-time thermal rate calculated based on traditional idealized steady-state conditions exhibits considerable fluctuation due to the complex factors inherent in dynamic processes

rendering it ineffective for making accurate judgments about unit operational economy. Addressing this issue

this paper focuses on a 600MW condensing extraction unit as the research subject

utilizing the APROS simulation software platform to establish a dynamic model for conducting load increase/decrease simulation experiments. By comparing the calculated thermal rates under different load change rates with the results from the steady-state model

the dynamic characteristics of the thermal rate during the unit"s load-changing process are analyzed. This paper proposes a method to correct the dynamic thermal rate

dividing the real-time thermal rate into steady-state thermal rate and dynamic thermal rate deviation. Furthermore

based on the Extra Trees algorithm

a predictive model for dynamic thermal rate deviation is established. With the steady-state thermal rate serving as the indicator for the thermal economic evaluation of the unit

this study offers a novel approach for assessing the economic performance of thermal power units during dynamic processes.