Unified Algorithm for Multilevel Inverters

Abstract

Multilevel inverters (MLIs) are broadly classified into symmetrical and asymmetrical types based on the DC-link voltage distribution. Asymmetrical MLIs are further classified into binary, quasi-linear and ternary hybrid inverters based on the geometric progression of the DC-link voltages. This study focuses on a new type of random asymmetrical MLIs (RASMLIs) where the DC links voltages are random and randomness is not pertained to any specific geometric progression. This study proposes an optimised vector search algorithm which is unified for symmetrical, asymmetrical and RASMLI. The concept of gradient-intercept equation and its corresponding Gs is the basis of the algorithm for identifying the proximal vectors. The algorithm determines the maximum DC bus utilisation at which the inverter can effectively produce quality output with minimal total harmonic distortion. As the redundant vectors reduce due to randomness the algorithm also aims at reducing the commonmode voltage with minimum switch transition. To validate the proposed algorithm, a T-type converter with H-bridge modules cascaded along each phase is considered. An experimental prototype of the same was developed and random DC voltages were fed using isolated DC supplies with induction motor as load. Nexys-4 field programmable gate array was used to implement the algorithm.

Behavior of a Dual Stator Induction Machine Fed by Neutral Point Clamped Multilevel Inverter

“…One of the main objectives of the multiphase machines and the reliability are known; for example, in the case of multistar machinery for failure in a phase we just need to disconnect the star that contains the faulty phase, which reduces torque, deployed 50% with a significant increase in torque ripples and magnetic Song [5][6][7][8]. Through the history of multiphase machines, the most popular multiphase machine type has been a dual three-phase machine.…”