Growing Neural Gas based MPPT of variable pitch wind generators with induction machines

Abstract

This paper proposes a maximum power point tracking (MPPT) technique for variable pitch wind generators with induction machines, which can suitably be adopted in both the maximum power range and the constant power range of the wind speed. To this aim, an MPPT technique based on the Growing Neural Gas (GNG) wind turbine surface identification and corresponding function inversion has been adopted here to cover also the situation of constant rated power region. This has been obtained, by including the blade pitch angle in the space of the data learnt by the GNG, and feeding back the estimated wind speed to compute the correct value of the pitch angle permitting the machine to work at rated power and torque. A further enhancement of the pitch angle selection by a simple Perturb & Observe (P&O) method has been integrated to cope with the wind estimation errors at machine rated speed. The proposed methodology has been verified both in numerical simulation and experimentally on a properly devised test set-up.Index Terms--wind generator, induction machine, maximum power point tracking (MPPT), neural networks (NN), variable pitch turbines NOMENCLATURE u sg u = u sgd +j u sgq : space vector of the grid side inverter voltages in the grid voltage reference frame; u s g = u sgD +j u sgD : space vector of the grid side inverter voltages in the fixed frame; u g u = u gd +j u gq : space vector of the grid voltages in the grid voltage reference frame; u g = u gD +j u gQ : space vector of the grid voltages in the fixed reference frame; i sg u = i sgd +j i sgq : space vector of the grid side inverter currents in the grid voltage reference frame; i sg = i sgD +j i sgQ : space vector of the grid side inverter currents in the fixed reference frame; ρ s = angle of the grid voltage space vector; u sgA ,u sgB ,u sgC : grid side inverter phase voltages; i sgA ,i sgB ,i sgC : grid side inverter phase currents; L, R : interconnection series inductance and its parasitic resistance; P, Q: active and reactive powers exchanged between the inverter and the grid; ω r = pulsation of the electrical grid; u sxy = u sx +j u sy :space vector of the machine side inverter voltages in the rotor flux reference frame; u s = u sD +j u sQ :space vector of the motor side inverter voltages in the stator reference frame; i sxy = i sx +j i sy :space vector of the motor side inverter currents in the rotor flux reference frame; i s = i sD +j i sQ :space vector of the motor side inverter currents in the stator reference frame; u sA ,u sB ,u sC : motor side inverter phase voltages; i sA ,i sB ,i sC : machine side inverter phase currents; t e = electromagnetic torque; ψ r = rotor flux linkage space vector; ρ r = angle of the rotor flux linkage space vector; p = number of pole pairs; ω mr = machine speed (in mechanical angles); ω rm = electric speed of the rotor flux linkage; ω T = wind turbine speed; P m = turbine mechanical power; P = generated electric active power; Q = generated electric reactive power; C p = performance power coefficient of the turbine; C T = perf...

Overall Description of Wind Power

“…In [34], [38] and [39] neural networks are used to assess the TSR, in [40] growing neural gas learning method is used to improve MPPT controller, while in [41] and [42] TSR method based on fuzzy logic is presented.…”