Genetic algorithm based induction machine characterization with application to adaptive maximum torque per amp control
Abstract
There has been considerable research in developing improved induction motor models. One recently developed model simultaneously includes magnetizing path saturation, leakage saturation, and a highly flexible transfer function approach to represent the rotor circuits. This alternate QD model (AQDM) is also computationally efficient in that it is non-iterative at each time step. It is considerably more accurate than the classical QD model (CQDM). However, the suggested characterization procedure available in the literature is complicated and time consuming. In this work, a new simplified characterization procedure for the AQDM is proposed. The proposed procedure employs a genetic algorithm as an optimization engine to identify the parameters of the AQDM by simultaneously considering per-phase fundamental frequency impedance and stand-still frequency response (SSFR) impedance. The proposed approach is validated by comparison of current ripple predictions and by application to maximum torque per ampere (MTPA) control design. Based on the AQDM characterized by the proposed approach, the MTPA control strategy is shown to achieve a desired torque with the minimum possible stator current. This is favorable in terms of inverter operation and nearly optimal in terms of machine efficiency. However, this work demonstrates that this MTPA controls perform sub-optimally as temperature varies. An adaptive MTPA control strategy is set forth that always achieves optimal performance regardless of rotor temperature, and does so without exhibiting hunting phenomenon. In addition, since the performance of the proposed adaptive MTPA control strategy relies significantly on the accuracy of rotor resistance, a new on-line rotor resistance estimator based on AQDM is proposed to predict the rotor resistance in good accuracy regardless of changing operating conditions. Finally, computer simulation and laboratory experiments are provided to validate the performance of both the proposed adaptive MTPA control strategy and the on-line rotor resistance estimator.
Degree
Ph.D.
Advisors
Sudhoff, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.