Under the condition of a high penetration rate of photovoltaic grid connection in medium and low voltage distribution networks

the existing photovoltaic control methods have the problems of difficulty in accurately describing the dynamic safety boundary of photovoltaic

a lack of effective coordination resulting in limited control level

insufficient control timeliness under rapid source and load fluctuations

and lack of feedback in execution. In view of this

this paper proposes a high penetration distributed photovoltaic feeder-station closed-loop control strategy based on a dynamic operation domain and an improved NSGA-III algorithm

which calculates aggregated control instructions in the feeder-station and accurately decomposes them to the low voltage side photovoltaic inverter for execution

forming a complete closed-loop control link with multi-level coordination. Firstly

a feeder-station closed-loop control framework with the whole process of "down-up-down" is constructed to realize the aggregation

centralized scheduling

and distribution execution of the distributed photovoltaic operation domain; secondly

a multi-stage distributed photovoltaic multi-objective optimization model considering feeder-station constraints is proposed to accurately describe the network topology of medium and low voltage systems; finally

the third generation non-dominated sorting genetic algorithm (NSGA-III) is improved based on the Latin hypercube sampling (LHS) method and the historical information reference point update strategy. The optimization model is solved in sequence to achieve efficient control. Through two real substation scenarios at different scales

it is demonstrated that the proposed method can more effectively and accurately characterize the dynamic operation domain of photovoltaics

enable closed-loop precise control of medium- and low-voltage levels

and fully leverage the potential of distributed photovoltaic active/reactive support.