Sensors-less neural maximum power point tracking control of induction machines wind generators by growing neural gas and minor component analysis EXIN + reduced order observer

Subject of this work is a maximum power point tracking technique for high-performance wind generator with induction machine based on the growing neural gas (GNG) network and the minor component analysis (MCA) EXIN + neuron. The main idea is to create a fully sensors-less system, meaning a system neither with the wind speed sensors nor the machine speed sensor. The GNG network has been used, trained off-line, to learn the turbine direct characteristic surface torque against wind speed and machine speed and implemented on-line, exploiting the function inversion capability of the GNG, to obtain the wind tangential speed on the basis of the estimated torque and measured machine speed. The machine reference speed is then computed on the basis of the optimal tip speed ratio. With regard to the power conversion stage, a back-to-back configuration with two insulated gate bipolar transistor (IGBT) voltage source inverters has been chosen, one on the machine side and the other on the grid side. The field-oriented control of the machine has been integrated with an intelligent sensorless technique, the so-called MCA EXIN + reduced order observer. The performance of the adopted technique has been verified experimentally on a suitably devised test setup.