Automatic Self-Commissioning for Secondary-Saliencies Decoupling in Sensorless-Controlled AC Machines Using Structured Neural Networks

The focus of this paper is secondary-saliency decoupling in carrier signal injection-based sensorless control of AC machines using structured neural networks. Structured neural networks are utilized for automatic commissioning and decoupling of secondary saliencies including saturation-induced saliencies. Automatic commissioning process is necessary for easy implementation and for acceptance of the carrier signal injection-based sensorless control by drives industry. In comparison with classical compensation methods, such as lookup tables, this technique has advantages of reducing commissioning time and automating the process. These advantages are result of a physics-based design of structured neural networks, which is responsible for their scalability, and moderate size and complexity. In comparison with traditional neural networks, structured neural networks are simpler, physically insightful, less computationally intensive and easier to train.