| There is a growing need for sinusoidal inverters that can supply power to nonlinear loads and still maintain precise regulation of the output voltage magnitude, phase and total harmonic distortion (THD). Also, with the ever-increasing demand for power, UPS systems may have to be paralleled to form a multi-inverter system. A multi-inverter system can have many desirable features such as N + X redundancy operation, improved reliability, modularity and expandability.; This thesis studies the control methodologies that enable inverter systems to deal with nonlinear loads and be paralleled easily with high performance. Starting with a brief introduction to the basics of sinusoidal PWM inverter in Chapter 2, neural network (NN) control of the inverters is introduced in Chapter 3. Benefited from abilities of nonlinear functional mapping and fault tolerance, simulation and experiment results show superior performance of the proposed NN controller compared with a traditional PI controller. In Chapter 4 and 5, parallel operation of multi-inverters is discussed. An instantaneous average-current-sharing scheme is modeled and analyzed in Chapter 4. Based on the model established, some key issues of the parallel multi-inverter system are discussed, including stability of the current-sharing controller, impedance characteristics and voltage regulation. A full-state feedback controller for parallel multi-inverter systems based on optimal control methodology is presented in Chapter 5. Having a global view of the whole system, the proposed full-state feedback controller is easy to design. The robustness of the controller to the change of number of parallel inverters has also been investigated. The hardware implementation of the controller is simple. All modeling and theoretical analyses are verified by simulation and experiment results. Finally, suggestions for future research are addressed. |
