An advanced performance-based method for soft and abrupt fault diagnosis of industrial gas turbines

Integrating gas turbines with intermittent renewable energy must operate for prolonged periods under transient conditions. Existing research on fault diagnosis in such systems has concentrated on the primary rotating components in steady-state conditions. There is a gap in investigating the interplay between shaft bearing failure and performance metrics, as well as fault identification under transient conditions. This study aims to identify faults not only in the main rotating components but also in the shaft bearings under transient conditions. Firstly, the performance model and fault propagation model of gas turbines are established, and the influence of bearing fault on the whole engine performance is analysed. Then, the fault diagnosis method is determined and the dynamic effects are compensated in fault identification at each time interval. Finally, the steady-state and transient fault diagnosis are carried out considering the constant and sudden faults for the main rotating components and bearings. The average run time and maximum error during the engine life cycle are 0.1064 s and 0.0086%. For the proposed dynamic effects compensation method, the average computation time and peak error at every moment are 0.1152 s and 0.0143%, clearly superior to the benchmark method. These results provide evidence that the proposed method can correctly diagnose the fault of the main rotating components and shaft bearings under transient conditions. Therefore, the findings mark an advancement in real-time fault diagnostic techniques, ultimately enhancing engine availability while upholding secure and affordable energy production.