Parameters identification and power conversion of a photovoltaic system

In recent years, the use of photovoltaic (PV) energy has drawn much attention. In the year 2013, the total installed capacity reached 134 GW and grid-connected PV system has dominated off-grid PV system in the past few years. The increase in PV capacity leads to the research on better circuit topologies, more efficient maximum power point trackers and optimizing the whole PV system. To achieve these objectives, an accurate PV model that can replicate the output characteristic of PV modules using circuit simulator is required. The key challenge is to identify the unknown parameters of the model based on limited information from the manufacturer’s datasheet. This is addressed in this thesis by proposing a novel identification method based on particle swarm optimization with an inverse barrier constraint function with the temperature effect incorporated. The proposed method has been validated using four PV modules of different PV cell technologies. The results show that the errors of the absolute maximum power point’s power and maximum power point’s voltage are less than 0.02 % and 0.3 % respectively. For the copper indium diselende cell technology, it exhibits higher error at elevated temperature as compared to the crystalline cell technologies. This proves that the single diode PV model is unable to properly model the output characteristic of this particular PV cell technology. In addition, the emerging of other PV cell technologies such as amorphous silicon and multi-junction gallium arsenide that has different cell characteristic also results in a need for new PV modeling approach. To fill up this research gap, a generalized multi-dimension diode PV model is proposed that allows greater flexibility to replicate the characteristics of various new PV technologies. The proposed generalized PV model is able to represent the existing PV models in the literature and provides larger current-voltage (I-V) coverage. By having larger I-V coverage, it can match a wider range of I-V curves of different cell technologies under various irradiances and temperatures. The proposed PV model has been benchmarked with the single diode model and double diode model. Based on the mean root-mean-square error, the proposed method shows an improvement of 9.33 % and 50.8 % as compared to the single diode model and double diode model respectively. The proposed PV model has been used to determine the optimal PV models for 13 different PV modules made of various PV cell technologies. The optimal PV models are determined based on their lowest maximum power point error and lowest root-mean-square error. The output from a PV system requires an inverter to convert it into ac power for the grid. A new inverter named sigma-Z-source inverter has been developed in this research. In contrast to existing transformer based Z-source inverter, the new inverter improves the voltage gain by reducing the turn ratio. This leads to higher modulation index which lowers the voltage stress across the semiconductor devices. Moreover, a transformer with a lower leakage inductance can be used which reduces the voltage drop across the dc-link. In a comparative study conducted between TZ-source inverter and the proposed sigma-Z-source inverter, it is found that the proposed topology uses less winding than the TZ-source inverter. This reduces the size of the transformer for up to 30 % for a dc-ac voltage gain of 3. From the experimental results, it is observed that by reducing the turn ratio from 2 to 1.5, the semiconductor stress is reduced by 6.56 % and the voltage drop across the dc-link is reduced by 56.62 %. With this feature of high voltage gain with reduced size, the proposed inverter is an excellent candidate for the ac-modular PV power system scheme which is the most promising future configuration for a PV power system.Doctor of Philosophy (EEE