Session:

GSW - Graduate Student Workshop

Title:

Well-Separated Switching Instances for Approximating theOptimal PWM using Quantum-Inspired Evolutionary Algorithm

Authors:

Shiyan Hu

Abstract:

The selected harmonic elimination (SHE) pulse-width modulation (PWM) aims to select the switching instances (angles) in such a way that a waveform with a particular characteristic is obtained and a certain criterion is minimized. The algorithms in the literature so far do not give consideration for separation of the consecutive switching angles computed from the equations, which is however a very important issue for practical applications. In this paper, a new algorithm is proposed to solve this nonlinear problem under the constraint that any two consecutive solutions are well separated from each other. The algorithm first transforms the nonlinear equations to a polynomial problem, then uses the Quantum-inspired Evolutionary Algorithm (QEA) to find the roots with the necessary amount of separation. Other than the standard QEA, our QEA favors localized search which is more suitable in our case. Since the obtained switching angles are reasonably distant from each other, they can be directly applied for inverters to alter directions without manually adjusting the angles as with other methods. Essentially, our method computes the best possible tradeoff between the maximum error for system performance and the minimum distance between consecutive switching angles. The polynomial is ill-conditioned and our algorithm is robust.