For parallel operation of large photovoltaic power station, connection impedance is different because of different location of each power generation unit, and operation status of the maximum output photovoltaic power of generation units are inconsistent. They generate difficulty on the interconnection conditions of photovoltaic power plant scheduling and control. If deviating from the parallel conditions, the electromagnetic circulation will be generated between each photovoltaic unit, which will cause great loss and harm to the operation of the equipment, and affect the operation efficiency. Recently, some methods were presented to control the grid connection of photovoltaic power generation, which focus on maintaining the voltage level of the grid connection point, but most of them suffered from eliminating the differences of wiring impedance and operation parameters among the power generation units in the station. To address these issues, this paper establishes a mathematical model of reactive power optimization for time-varying tracking of maximum output power of grid-connected photovoltaic power station, and gives a fast solution method. By quickly solving the control base point of inverter, the purpose of tracking the maximum grid-connected output of photovoltaic power station with time is achieved. Firstly, taking the output active power and reactive power of photovoltaic inverter as decision variables, a mathematical optimization model is established to maximize the grid-connected power of the whole photovoltaic power plant under the constraints of power flow, voltage level and photovoltaic power output. The state change of photovoltaic grid-connected system is small in very short time interval, so the nonlinear optimization model is linearized by Taylor expansion and linear fitting. Through sensitivity calculation, the operating basis point of the reactive power of the inverter is quickly solved, and the light will be automatically discarded when necessary. This model uses the concept of sensitivity to solve the control point based on the online optimization iterative process, thus leading to a computationally fast model. Simulation results show that, if two photovoltaic power generation units operate in parallel, and the photovoltaic output active power has little difference, the inverter output reactive power is approximately proportional to the line impedance distribution to maintain grid voltage consistency. Using illumination data from 00:00 to 24:00 and compared with the maximum power point tracking (MPPT) output, the results show that, during the period of strong illumination (10:00-15:00), this method automatically abandon light, reduce the active power output of photovoltaic power generation unit, and ensure the voltage meets the constraint by giving way of active power to reactive power. At the same time, the voltage control effect analysis also verifies this point. Before the algorithm in this paper is adopted, the photovoltaic power supply is high during 11:00-17:00, making the node voltage exceed the upper limit. While the node voltage can be controlled within the allowable range by using this algorithm. When the node voltage exceeds the upper or lower limit due to the high or less photovoltaic power generation, the inverter will absorb or emit reactive power to ensure the voltage within limit. The following conclusions can be drawn from the simulation analysis: ①Compared with the traditional nonlinear optimization, the proposed method significantly reduces the solution time and improves the computational speed. The model uses the fast regulation of the inverter to quickly solve the active and reactive power operating base points through online optimization, and discards light to ensure the grid-connected voltage when necessary. So the maximum output power of the photovoltaic power station can be tracked time-varying. The model aims at maximizing the grid-connected power of PV and indirectly seeks to minimize the system loss. By regulating the output power in real time through the inverter, the active power loss is significantly reduced compared with the operation of the power station before this method is adopted. © 2023 Chinese Machine Press. All rights reserved.
