| Harmonics are a damaging phenomenon in power systems. They can lead to excessive heat and noise in the loads and create large amount of energy losses either in transmission systems or distribution systems. Various frequency harmonics may resonate with network components, like capacitors and transmission lines, which will enlarge the harmonic energies and increase their impacts. Due to the negative impacts generated by harmonics, many scholars have engaged in researching this issue and many strategies were developed to handle power system harmonic problems. These schemes are either active mode or passive mode. As a kind of active mitigation strategy, possessing the features of flexibility and high efficiency, PWM controlled power electronic devices have obtained more and more recognition.; Selected harmonic elimination PWM technique is a special kind of optimal PWM technique. It can generate high quality output waveform through elimination of specific lower order harmonics. The inverter switching angles can be determined by solving a set of nonlinear equations. These transcendental equations are derived from the signal frequency domain analyses and have disadvantages in surplus harmonics distribution in the output spectrum. They are also heavily ill-conditioned when the number of eliminated harmonics becomes large. Meanwhile, the roots search with traditional algorithm depends on the iteration initial values and can only be used in off-line situations because of the low calculation speed.; A novel algorithm is introduced in this work for the most general case of a single-phase inverter bridge. It successful transforms the original non-linear equations to a set of polynomials and then applies simplex homotopic fixed-point algorithm for a parallel search of the polynomial roots. It has high accuracy, fast speed and is robust. Furthermore, it does not suffer from error propagation. With typical contemporary digital signal processing (DSP) chips, this algorithm can be used to set amplitudes of several tens of harmonics within one cycle of the ac power waveform. Since the roots are calculated independently from each other, the algorithm can be easily implemented for parallel processing. An optional eigenvalue algorithm for ill-conditioned problem polynomial roots finding has been also introduced.; To overcome the disadvantages of the original SHE-PWM technique, several optimal harmonic control techniques are presented in this dissertation. Applying these techniques, we can reasonably control the crest of surplus harmonics. On the other hand, the best possible tradeoff between the maximum error in the system and the minimum distance between consecutive switching angles can be obtained. By optimizing the switching angles distribution, we can flexibly re-distribute the energy of harmonics and the output spectrum to obtain a high quality waveform. |
