Distributed Predictive Control Based On Data-Driven Subspace Approach And Its Application In Grid-Connected Solar PV Generation

With the increasingly pollution of the environment, the demand for clean energy is growing, such as solar energy, wind energy. With the growing scale of solar photovoltaic (PV) generation system, the core research and application of grid-connected solar PV generation system is force on the structure of PV generation system and the output power quality impact on the grid. The grid-connected inverter is the most important part of grid-connected solar PV generation system. The control performance of inverters is essential to the efficiency of the output currents> the security and stability of the system.In this paper, a distributed predictive controller based on data-driven subspace approach is applied to the distributed solar PV generation system. The data-driven control approach uses the input and output measurements of grid-connected PV generation system to calculate the system parameter matrixes and control the real-time currents of the system. The proposed method overcomes the drawbacks of traditional modeling process, which is complex and time-consuming. The performance of the whole has been improved by using the distributed control method.In this paper, firstly, the description of the data-driven control approach, predictive control method and the distributed predictive control method were given. The present situation and application of distributed generation were introduced. Next, through the research of the structure and control method of the inverters in grid-connected distributed PV generation system, the control objection was given. Then the model method of traditional predictive method and the data-driven subspace approach was described in detail. Also, the detail description of the predictive control method based on data-driven approach was given. Finally, the simulation model of the grid-connected solar PV generation system is established in Matlab/Simulink by using the SimPowerSystems. Then a centralization predictive controller based on data-driven approach and a distributed predictive controller based on data-driven approach were designed to control the system. The simulation illustrates that the quality of subsystem output currents in grid-connected solar PV generation system was conformed to the actual grid connected standards and the designed controller has a good dynamic performance and robustness.