For high performance motor controls applications such as electric vehicles, accurate motor parameter knowledge is required. Motor parameters like inductances, resistance and permanent magnet flux linkage are difficult to obtain and measure directly. These parameters vary considerably depending on operating conditions. Various methods are available in literature to obtain motor parameters offline. Usually motor designers use finite element analysis to calculate inductance estimate. Normally motor data necessary to come up with finite element model is unavailable and not provided by manufacturers. Even if provided, simulation results rarely match experimental results. Other methods commonly used in industry are locked rotor test and no-load test. However, parameters obtained by using such tests differ from parameters in real operating conditions. In some other techniques, machine inductances are calculated by running the machine at constant speed and measuring voltage, current and speed thus requiring special measuring instruments and sensors which are not always available. In this paper, we proposed a new scheme to approximate d-axis and q-axis flux linkage using measured torque, dq-axis measured current, and dq-axis voltage commands to the inverter. d-axis and q-axis flux linkages are estimated for different sets of d-axis and q-axis currents. In our approach, the parameter identification problem is converted to an optimization problem. dq-axis flux linkages are estimated using an artificial intelligence based method. The results obtained show that parameters can be accurately estimated and can be used to predict optimal dq-axis current trajectory for various types of motor control algorithms.