An online technique for condition monitoring the induction generators used in wind and marine turbines

Induction generators have been successfully applied to a variety of industries. However, their operation and maintenance in renewable wind and marine energy industries still face challenges due to harsh environments, limited access to site and relevant reliability issues. Hence, further enhancing their condition monitoring is regarded as one of the essential measures for improving their availability. To date, much effort has been made to monitor induction motors, which can be equally applied to monitoring induction generators. However, the achieved techniques still have constrains in particular when dealing with the condition monitoring problems in wind and marine turbine generators. For example, physical measurements of partial discharge, noise and temperature have been widely applied to monitoring induction machinery. They are simple and cost-effective, but unable to be used for fault diagnosis. The spectral analysis of vibration and stator current signals is also a mature technique popularly used in motor/generator condition monitoring practice. However, it often requires sufficient expertise for data interpretation, and significant pre-knowledge about the machines and their components. In particular in renewable wind and marine industries, the condition monitoring results are usually coupled with load variations, which further increases the difficulty of obtaining a reliable condition monitoring result. In view of these issues, a new condition monitoring technique is developed in this paper dedicated for wind and marine turbine generators. It is simple, informative and less load-dependent thus more reliable to deal with the online motor/generator condition monitoring problems under varying loading conditions. The technique has been verified through both simulated and practical experiments. It has been shown that with the aid of the proposed technique, not only the electrical faults but also the shaft unbalance occurring in the generator become detectable despite the external loading conditions. Moreover, the rotor and stator winding faults can be readily discriminated through observing the variation tendencies of the proposed condition monitoring criteria.

► The technique is informative and reliable in comparison of existing methods.
► The criteria are effective in detecting both mechanical and electrical faults.
► Stator and rotor winding faults can be discriminated with the aid of the criteria.
► The proposed criteria are simple in calculation, thus suited for online use.