| Energy crisis and climate change is one of the major challenges facing the world.The solar photovoltaic industry came into being as humans efforts to address thesechallenges. As the interface device between solar photovoltaic array and the utility grid,photovoltaic grid-connected inverter is the core component of the grid-connectedphotovoltaic system. Photovoltaic grid-connected inverter plays a crucial role in thedevelopment and utilization of new energy. It is important to study the photovoltaicgrid-connected inverter. Establishment of mathematical model for photovoltaicgrid-connected inverter, is the basis of all the research work. However, the mathematicalmodel provided by the manufacturer of power inverters is based on analyticalapproaches and contains a large number of simplifying assumptions. Thus, usersexperience difficulty in judging its accuracy and have had to revise the model throughthe product validation test one by one. This restricts the study of modeling, simulation,operation control, power quality and other aspects of research and design in the newpower system with distributed generation. Hence, a new distributed power invertermodeling method is urgently needed for theoretical analysis and construction of agrid-tied new energy power generation system.In view of the photovoltaic grid-connected inverter modeling problem, based onthe nonlinear system identification modeling method, the different photovoltaicgrid-tied inverters modeling problems had been deeply studied, including single-phase,three-phase photovoltaic grid-connected inverters identification modeling methods andthree-phase photovoltaic inverter model simulation.The system identification method of single-phase photovoltaic grid-connectedinverter NARX model was proposed. For the black box feature of commercialphotovoltaic grid-tied inverters, as well as the strongly nonlinear problem of the inverterwhich can not be solved by existing linear modeling approach, in this approach, theinverter was considered as a black box, wherein it was not necessary to know thetopology and the parameters of the inverter internal circuits and power switchingdevices, as well as the type and logical relations of the control system. It only used theinput-output external measurement data of the inverter, based on the NARX modelnonlinear system identification techniques, to create an accurate mathematical model.The model can accurately imitate the behavior of the commercial inverter and has simple structure, small amount of computation. It takes a good balance between thecomplexity of the model and the model accuracy. It is suitable for power system withthe grid-connected photovoltaic system scheduling, joint operation and coordinatedcontrol, stochastic simulation research areas, in which the fast modeling and simplemodel structure are required.Then the model identification method of single-phase photovoltaic grid-tiedinverter based on the Wiener Model was proposed in the point of further improving thethe accuracy of identification model. Based on nonlinear theory and discrete-timemethod, the single-phase photovoltaic grid-connected inverter was considered as a blackbox and was modeled only using input-output information of the inverter under differenttypical light conditions during multiple time periods. Finally, in the same experimentalconditions, the comparative analysis was conducted among the single-phasephotovoltaic grid-tied inverter Wiener model, the Hammerstein-Wiener model and theNARX model which was proposed in the second chapter. The results show that thesingle-phase photovoltaic grid-connected inverter Wiener model and its identificationmodeling method can better reflect the dynamic behavior of the actual single-phasephotovoltaic inverter under different weather conditions, and with higher accuracy andadaptability. However, the Wiener model is relatively complex and time consuming. It issuitable for applications requiring the precise model of photovoltaic grid-connectedinverter.Photovoltaic grid-connected modes include single-phase and three-phase mode.When using identification modeling method for three-phase inverter, the input andoutput variables increases, the model becomes complicated, and the model parameter isdifficult to identify. Based on the identification modeling approach of the single-phasephotovoltaic gird-connected inverter, the nonlinear system identification modelingmethod for three-phase photovoltaic grid-tied inverter has been proposed. In order tosimplify the modeling of three-phase inverter, three-phase system has been simplified.At the input, the input voltage and current were selected as the inputs. At the output,drawing on the symmetrical three-phase circuit analysis methods, one phase active andreactive power outputs were optionally chosen for the identification and reconstructionbased on a three-phase symmetrical relationship. The identification model wassimplified as a dual-input dual-output system, which reduced the difficulty ofidentification. Then the NARX model and Wiener model were respectively applied tobuild a three-phase photovoltaic gird-connected modeling approach after simplification. In order to analyze the pros and cons of different modeling methods, a dual-stage,three-phase, grid-connected photovoltaic system had been established. In the system,simulation studies had been conducted for a variety of three-phase photovoltaicgrid-tied inverter model which included the detailed model, the NARX model and theWiener model. The nonlinear system identification modeling method based on NARXmodel is optimally selected. It can be used as three-phase photovoltaic gird-connectedinverter identification modeling method, because of its simple model structure, smallamount of computation and the advantage of high accuracy of the model. |
