The main purpose of many industrial applications is including control techniques which combine simplicity, robustness, and efficient performance in order to obtain a speed response characterized by fast dynamic, small overshoot, and resistance to disturbances. This combination is achieved through the proposed structure in this paper, where the conventional proportional integral (PI) controller is replaced by a hybrid controller based on an intelligent approach namely fuzzy logic (FL), and on a PI anti-windup technique; the conventional switching table is replaced by the selector table based on artificial neural networks (ANN). The extended Kalman filter (EKF) software sensor is used to estimate the rotor speed and the flux components applied to direct torque control (DTC) of an induction motor (IM). This new intelligent control infrastructure optimizes and reduces the torque, stator flux and stator current ripples. Simulation tests prove the good performance of the proposed sensorless strategy in large operating conditions. Simulations are done using matlab/simulink environment.