Control and power management of photovoltaic systems with plug-in hybrid electric vehicles as energy storage

The distribution network has experienced an increasing level of photovoltaics (PVs) and plug-in hybrid vehicles (PHEVs) integration in recent years. Investigating the potential influence of integrating these sources into distribution networks is difficult and requires the development of a suitable system simulation model for a grid-tied PV system with energy storage. The major objective of this thesis is modelling, control and power management of a grid-connected PV system with PHEVs as energy storage. The parameters of the PV array and Tremblay's battery models integrated within the whole system simulation model are estimated using the parameter estimation techniques. The simulation models developed throughout the thesis are implemented in MATLAB/SimPowerSystems environment. Experimental testing of BP 380 PV modules is conducted to validate the effectiveness of the PV module model. A suitable control and charging strategy is also developed to control the charging and discharging processes of the PHEV battery. The major novelty of the work described in the thesis lies in three aspects: (1) parameter identification of PV modules using a genetic algorithm (GA) approach to improve the accuracy of the model parameters; (2) parameter identification of Tremblay's battery model using a novel quantum-behaved particle swarm optimization (QPSO) parameter estimation technique; (3) development of a charging strategy for PHEVs to optimally coordinate the power flow among the system based on the State of Charge (SOC) scenario of a day. This thesis begins with a study of modelling approaches for a PV cell and selects the single diode model (SDM) to model the PV array. A critical review of three parameter estimation techniques for the SDM is presented. A novel GA approach to parameter estimation for the SDM is also proposed. Simulation results are presented to show the advantages of the GA approach over Villalva's iterative method [1]. Experimental testing of a BP 380 PV module is conducted to validate the effectiveness of the SDM in modelling the experimental current-voltage (I-V) and P-V characteristics. A PV array simulation model is developed using the SDM calibrated through parameter estimation. Secondly, the thesis presents overviews of MPPT techniques, DC-DC converter topologies, and grid-tied PV inverter topologies for MPPT applications. A review of the input voltage control of DC-DC converters is presented, especially on the voltage mode control (VMC) and current mode control (CMC). This leads to the development of a proportional-integral-derivative (PID) controller. This controller is used to transform the PV array voltage tracking error into the duty cycle to control the operation of the boost converter interfacing the PV array. Four cases are developed based on whether the effect of the DC link capacitor is considered, and different linear models are selected for modelling a PV array. According to the assumptions given in the four cases, the small signal model of the boost converter is developed by adoption of the PWM model in each case. Consistently with the derived small signal model, four different control-to-input voltage transfer functions are developed and the corresponding parameter sets of the PID controller are determined. To investigate the performance of different parameter sets of the PID controller, a single-phase grid-tied PV system model is employed. Four case studies have been conducted to analyse the effects of different tuning settings of the PID controller on the PV array voltage response. The most appropriate PID controller parameter settings are selected leading to the fastest rise time and zero steady state error of the PV array voltage response among the four different cases. Thirdly, the thesis provides a review of existing parameter estimation techniques used to parameterize Tremblay's battery model for PHEV batteries. These existing techniques include: particle swarm optimisation (PSO), GA, and simulated annealing (SA). A QPSO is proposed in the thesis to estimate the model parameters of Tremblay's model and the resultant discharge curve is compared to those generated by GA and PSO approaches. The simulated battery discharge curves obtained from GA, PSO, and QPSO parameter estimation techniques are compared to the experimental data together with the simulated discharge curve obtained from Tremblay's parameter estimation method. Results of the comparison indicate that the QPSO parameter estimation technique converges to acceptable solutions with fewer iterations than the GA and the PSO techniques. The QPSO parameter estimation technique also needs less tuning effort than the GA and the PSO techniques since there is only one tuning parameter involved in the QPSO approach. Finally, this thesis also develops control and charging management strategies for controlling the charging operation of a PHEV battery. The PHEV battery can be charged from the PV array during the daytime when solar power is sufficient, and from the grid at night. The charging power is determined in such a way that the actual SOC of the PHEV battery follows a pre-set SOC reference when the PHEV battery is charging from the grid. The charging power reference is set to the power difference between the PV array and the local load when the PHEV battery is charging from the PV array. Results show that the charging management strategies can achieve the objectives of charging using the PV array power when solar energy is available, and using the grid power at night time.