Neural-network-based sensorless maximum wind energy capture with compensated power coefficient

This paper describes a small wind generation system where neural network principles are applied for wind speed estimation and robust control of maximum wind power extraction against potential drift of wind turbine power coefficient curve. The new control system will deliver maximum electric power to a customer with light weight, high efficiency, and high reliability without mechanical sensors. The concept has been developed and analyzed using a turbine directly driven permanent-magnet synchronous generator (PMSG). In addition, the proposed method is applied to a 15-kW variable-speed cage induction machine wind generation (CIWG) system. The simulation studies of a PMSG small wind generation system and experimental results of a CIWG are provided to verify the validity of the method.