A Waveform-Informed Fault Diagnosis Algorithm for Combined Inverters Under Variable Input Voltage Disturbances Based on VMSS

The on-board inverter of the more electric aircraft is powered by high-voltage direct current (HVDC) busbar. However, the voltage input to the inverter is frequently concomitant with uncertain disturbances that attributable to energy feedback from numerous sources and sudden alterations in high-power loads. Those disturbances further affect the open-circuit fault characteristics of the inverter, consequently introducing uncertainty in fault characterization, which can easily lead to misdiagnosis of faults. To solve the problem, a waveform-informed diagnostic algorithm based on the variable mass sinusoidal-wave structure (VMSS) is proposed in this paper. First, the concept of variable mass sinusoidal-wave structure of the inverter output signal is proposed to calculate the equivalent quality of the signal sequence. On this basis, variables such as quality factor and quality deviation factor are introduced to extract the peak-to-valley and half-period characteristics of the fault waveform to realize fault detection and primary classification. After that, four types of adjacent invariant sampling counting (AISC) variables are designed to reflect the differences of level signals under fault conditions by utilizing the level characteristics of internal signals under different faults. Finally, the results of the classification algorithm are combined to realize the localization of all single-switch and multi-switch faults in the inverter. The effectiveness and superiority of the algorithm is finally verified by experiment and comparison. Compared with existing algorithms, the proposed method effectively eliminates the impact of input voltage disturbances caused by HVDC busbar from the front end of the inverter on diagnostic results. Meanwhile, the proposed method diagnoses more failure modes with less hardware control cost, and does not depend on a specific modulation strategy.