Insulation State Assessment of Rotating Electrical Machines by Employing Generalized Additive Models

A growing body of literature has investigated the fault detection, diagnosis and monitoring methods of rotating electrical machines. However, there are still some critical issues, such as aging of electrical insulation. In practice, most of the stress variables responsible for insulation degradation, e.g., electrical, thermal, mechanical factors are not available or measurable, especially during operation. Recently computational intelligence approaches are being applied for that class of problems. However, the Generalized Additive Models (GAMs), due to their flexibility and lower data requirement and lower complexity, have gained attention. In this paper GAMs are applied to assess insulation state by mapping relationship between measurements, fault history and information from various sources with a varying degree of uncertainty. Results support that GAMs are promising candidates for online insulation assessment and fault diagnosis systems on account to their remarkable advantages.