In order to solve the problem of intelligent fault diagnosis of thermal system of thermal power unit, the application modelling of intelligent diagnosis of thermal cycle system is proposed. This paper first filters the influence of the change of valve opening of steam turbine on the performance index under the condition of constant power, and qualitatively analyzes the actual change of the performance index compared with the reference working condition when the component fails, so as to eliminate the mutual influence between the components. Then, the selection principle of thermodynamic parameters is determined, the irreversible loss in the structural theory of thermal economics is introduced as the performance index, and the model is used to quantitatively calculate the change of performance index of each component under fault conditions to diagnose the failed component. Finally, APROS simulation software is used to simulate various fault conditions of 330 MW units in a power plant. The experimental results show that the recognition accuracy of the monitoring system designed in this paper can reach 98.33%. In conclusion, the method in this paper proves the feasibility of intelligent fault diag­nosis of thermodynamic system network.