Optimal Elman Recurrent Neural Networks-Based Induction Motor Drives With Estimated Machine Parameters

A machine learning-based Elman recurrent neural network (ERNN) parameter estimator and optimized PI gain-based adaptive indirect field-oriented control (IFOC) strategy is implemented to reduce the improperly field orientation of an induction motor drives (IMDs). To avoid the slow convergence rate, falling into local minima, and provide a derivative-free solution, the nature-inspired metaheuristic algorithm is integrated into ERNN training. In this article, marine predators an optimization algorithm (MPOA) is employed to obtain the best-fitted predictor model. It plays a significant function in appropriately selecting network weights and biases. To enhance the IMD performance in industrial and automotive applications, an optimal ERNN is employed to estimate accurate electrical parameters (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r_s, r_r. l_s. l_r</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">,</sub> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">l_m</i>), and cost-effectual speed over a broad operating region during dynamic conditions. Further, a modified Space vector PWM is adopted for the switching operation of VSI-fed IMD to reduce the complexity by avoiding the dwell-time calculations. In addition to training of ERNN, MPOA is employed to evolve the optimized gains of a controller such as speed and current for the generation of reference signals through IFOC with minimized error. The system response shows that the proposed algorithm ensures effective robustness as well as stable operation during various dynamics.