Search

The effects of cross-coupled inductance are significant in high power density motors. It is observed here that the cross-coupled inductance causes current and voltage overshoots in the high-speed region. As the required voltage exceeds the voltage limit, the system becomes unstable. Here, the current dynamics are analyzed with a reduced model that incorporates a cross-coupled inductance, $L_{qd}$. Further, it is shown that $L_{qd}$ makes an uncanceled zero in the current transfer function, which invokes the overshoot. It can be compensated with an estimate $\hat{L}_{qd}$. However, it is not easy to obtain correct estimates of $L_{qd}$ in a wide current range. In this paper, an online overshoot suppression algorithm is proposed by using a high-pass filter. The simulation and experimental results provide some evidence for the effectiveness of the proposed algorithm. Finally, it is shown experimentally that the proposed algorithm cured the vehicle hunting problem after tapping off the gas pedal in a high-speed operation. [ABSTRACT FROM AUTHOR]

Coasting is a deceleration driving mode not recurring to an active braking while sustaining rolling as much as possible. In an electric vehicle (EV), coasting should be realized in theory by letting the $q$-axis current set to zero. However, unintentional braking takes place at high speeds when the gas pedal is tapped off. The reason is that negative torque is developed even when the q$-axis current is equal to zero. Such a braking reduces the coasting distance. To prevent undesirable braking, positive q$-axis current is often utilized in practice. In this paper, we focus on a braking torque component caused by a cross-coupling inductance . It is produced only by the $d$-axis current, and is normally small. However, its magnitude becomes noticeable in a deep field-weakening region. As a curing method, it is suggested to shift the angle of the reference frame clockwise. The magnitude of compensation angle is derived analytically using two different methods. The proposed compensation method was demonstrated by simulation and experiments. [ABSTRACT FROM AUTHOR]