Abstract: This research introduces a novel sensor-free control framework for three-phase asynchronous motors, employing an Adaptive Neuro-Fuzzy Inference System (ANFIS) to estimate rotor velocity. The proposed methodology utilizes easily accessible electrical parameters, namely stator voltage and current measurements, to eliminate mechanical speed transducers, consequently lowering implementation costs while improving system dependability. The ANFIS estimator is trained offline across multiple operational scenarios and achieves real-time speed prediction with exceptional precision (deviation < 2%) and substantial immunity to motor parameter fluctuations. The drive system implements Indirect Stator Flux Oriented Control (ISFOC), ensuring full independence between torque generation and flux regulation. For medium-to-high voltage industrial deployments, we employ a three-level Neutral-Point-Clamped (NPC) converter operated through Space Vector Modulation (SVM), which significantly lowers harmonic distortion and switching device voltage stress relative to traditional two-level designs. Extensive MATLAB/Simulink simulations confirm the integrated architecture's effectiveness, showcasing superior reference trajectory tracking, fast transient behavior, and robust performance during demanding scenarios such as speed reversals and load variations.

Keywords: ANFIS, Induction motor IM, ISFOC, SVM, three-level NPC inverter

Pages: 60-71