Compared with the conventional strategies for electrical drives such as Field Oriented Control (FOC), Predictive Current Control (PCC) shows the superiority of fast dynamic response and low switching frequency. Instead of an encoder, a speed-sensorless algorithm is applied to provide the rotor position for PCC, which reduces the hardware cost and complexity. However, the estimation accuracy is significantly penalized by the parameter deviation, which leads to unsatisfactory control performance like high torque and current ripples. To cope with this problem, an Extended Kalman Filter (EKF) is presented to estimate stator flux and rotor speed in this paper, which improves robustness to parameter variations and achieves high control performance. The steady-state, dynamic transient performance and robustness evaluation of the proposed scheme are experimentally validated on the 2.2kW induction machine (IM) platform. It is indicated that the proposed method shows excellent speed tracking ability and strong robustness to parameter deviations.