Model-free Subspace Control And Its Application On Grid-Connected PV Inverter

As the most important part of the solar grid-connected power system, the inverter is the interface between the solar panels and the grid, which can guarantee high quality of the power and high efficiency of energy conversion. Most of the inverter control strategies are model based, however after complex and consuming modeling process, the model may not be accurate, that’s why we study on model-free subspace control. And then subspace based H∞control was proposed, which was extended from subspace based LQG control. Finally subspace based LQG control and H∞control were applied to grid-connected PV inverter successfully.Firstly, the present situation and application prospects of subspace methods and solar power industry were introduced, and diverse control strategies of grid-connected inverter were studied respectively. The three-phase grid-connected inverter system was designed by studying on the solar power system.And then subspace predictive control and subspace based LQG control were introduced. Sometimes the system’s structure was complex, so it’s hard to form a model by mechanism modeling. Subspace predictive control not only combined system identification and control design, but also gave future predictive output values directly based on input-output data, avoiding the tedious model formation step, the predictor matrix can be one of the controller parameters, which can reduce the parameter number and simplify design process.A novel subspace based mixed sensitivity H∞control was proposed. While LQG can only handle white noise, but in reality we should consider interference signal sets, so extend subspace predictive control from LQG control to H∞Control, which can take uncertain in consideration. At first the best least squares subspace predictor was derived from available experimental data, then with simple matrix operations to get the H∞control law, no longer need solving complex Riccatti equations or linear matrix inequalities, It greatly reduced plant model order and the effects of noise in the measured data.Finally, subspace based LQG control and H∞control were applied to three-phase grid-connected inverter control. Subspace model control methods can not only greatly reduce the computational complexity, but also achieve good control performance. A grid-connected PV inverter model was established by Matlab/Simulink SimPowerSystems toolbox, and subspace-based LQG controller and H∞controller were applied to the inverter respectively. The simulation results show that the way which applied subspace predictive control to the grid-connected inverter is feasible, the output current of the inverter measure up to grid-connected standards, and has good dynamic performance and robustness.