In servo systems, the bandwidth of the current loop is increased by raising the switching frequency. However, the dead-time nonlinearity of the voltage source inverter (VSI) intensifies with the increase of switching frequency, causing the deviation between the actual and the theoretical output voltage, resulting in serious distortion of the inverter output current waveform.

The requirement of computational power constrains the implementation of high switching frequency control. Consequently, the dead-time nonlinearity compensation strategy should be more straightforward to decrease computational time, especially for high switching frequency applications. The amplitude increase is relatively modest at a low switching frequency but significantly surges at a high one, bringing on a severe degradation of the linear modulation region. It is an imperfect solution to compensate after the occurrence of dead-time, which unavoidably introduces compensation errors. Furthermore, the accurate solution of inverter nonlinear voltage error (INVE) under different currents represents a crucial aspect of achieving exact INVE compensation. Nevertheless, the existing methods are complex.

This paper proposes a novel strategy that combines no-dead-time double modulation wave pulse-width modulation (PWM) and inverter nonlinearity compensation for analyzing the dead-time nonlinearity and the nonideal characteristics of the inverter on the output voltage error. Firstly, according to the continuous current characteristics of the anti-parallel diode, the drive vacancy area is added between the complementary drive pulses to avoid the introduction of dead time. Compared with the ideal space vector pulse width modulation (SVPWM), an auxiliary modulating wave is added. Depending on the current polarity, its amplitude is adjusted up or down from the original modulating waveform. The underlap periods are generated between the complementary drive pulse by contrasting the double-modulating and the triangular carrier waves. It is practical to avoid both the bridge arm shoot-through and the introduction of dead time. Most importantly, the actual output voltage of this method is identical to the optimal voltage, which directly eliminates the dead-time nonlinearity and removes the limitation of dead time on the output duty cycle at a high switching frequency. Moreover, the control signals of each switching device are obtained based on the comparison between the double-modulating and the carrier wave. No additional control loop calculations are required, while the generated PWM signals are symmetric about the carrier midpoint.

Secondly, the inverter nonlinearity is equated to the INVE, which varies with the current. When the motor is at a standstill of ${{\theta }_{\text{e}}}={{0}^{\circ }}$, by injecting the ramp current signal into the direct axis and applying Kirchhoff's voltage law, the sum of INVE containing the nonlinear factor of the two-phase VSI is obtained. Finally, the relationship between the INVE and the current amplitude is calculated using the linear iterative interpolation approach, and the online compensation of the INVE is achieved.

The results show that the proposed strategy can increase the linear modulation region of the output voltage, eliminate the output duty cycle limitation derived from the dead-time, and effectively suppress the current harmonic distortion phenomenon caused by the dead-time nonlinearity of high switching frequency inverters. In addition, the strategy is easy to implement without additional control loop calculations, which can be applied to servo drive control systems requiring high switching frequencies.