Online adaptive SHE algorithm for multilevel converters and implementation with embedded control system

Abstract

Selective harmonic elimination pulse-width modulation (SHE-PWM) method has been widely used in various industrial multilevel converters befitting from low switching frequencies and precise elimination of low-order harmonics. Most of existing studies focus on accelerating calculation of switching angles or optimizing predictive neural networks. However, these offline methods cannot dynamically deal with the errors introduced by nonuniform dc sources of multilevel converters or biased placements of switching angles. This paper proposes an online adaptive SHE (OA-SHE) algorithm to compensate these errors. Errors caused by nonideal dc (NID) sources and leading-lagging placements (LLPs) are analyzed in detail. The error of output waveform is comprehensive result of NID sources and LLPs. By introducing feedback, the output waveform accuracy of multilevel converters can be increased greatly. Realization of the particle swarm optimization (PSO) with an embedded control system is presented to acquire good real-time performance. With the OA-SHE method, the value of particles has been higher than the initial value. High order harmonics, including 1st, 5th, 7th, 11th, and 13th components, are suppressed by the low-pass filter. Comparing to existing studies, this online algorithm does not need detailed model parameters and can be directly applied to practical applications achieving multi-objective optimization. Finally, both simulations and experiments are carried out to verify the effectiveness of this algorithm.