A novel method for fault diagnosis of hydro generator based on NOFRFs

• Modeling the hydro generator with non-linear output frequency response functions.
• A novel online identification method is proposed for NOFRFs of hydro generator.
• The nonlinear behavior of different states of hydro generator is analyzed.
• Some failure mechanism of hydro generator is classified.

Fault diagnosis and recognition of hydro generator are important issues which encounter in repair and security posture assessment. It influences the operational planning and security directly. Volterra series as an effective modeling method has been widely used in modeling and fault diagnosis of hydro generator, but the larger number of kernels limits its application in faults diagnosis. Non-linear output frequency response functions (NOFRFs) as a transformation style of Volterra series have a more intuitive visual and simple structure. In this paper, NOFRFs have been proposed to be employed in fault diagnosis of hydro generator, and a novel online identification method is proposed at the same time. Firstly, NOFRFs models of hydro generators are built. Secondly, a new method for online identification is proposed according to the operational characteristics of hydro generators. Finally, simulation verification has been done to demonstrate the advantage of the proposed method, and experimental studies are put forward on a hydro generator to analyze the failure mechanism in different states. The results indicate that the proposed method is useful and concise for fault diagnosis and nonlinear analysis of hydro generators.