摘要：AbstractThis paper contributes to efficient operation of pump-turbines under partial load conditions. The authors propose an adaptive perturb & observe (P&O) algorithm to search for maximum possible efficiency for a particular operating load. The adaptive P&O algorithm adapts the step size with the help of a fuzzy inference system. A pump-turbine and a doubly-fed induction machine (DFIM) have been modelled using dynamic velocity triangles, considering other dynamic losses and penstock dynamics and circuit dynamics, respectively. The DFIM has been used to vary the speed of the turbine, and speed reference tracking was accomplished with the help of the rotor side converter (RSC) of the DFIM. Initially PI controllers were designed for speed reference tracking. However, the paper also proposes a PI controller augmented with an artificial neural network (ANN) for speed controller. Results indicate: a) enhanced efficiencies for partial load conditions, b) reduced oscillations at the optimum speed, and c) enhanced speed tracking performance; efficiencies greater than 90% have been achieved for partial load conditions.
