In order to adapt to the development trend of large-scale access of distributed resources and loads as well as power market reform

it has gradually become the industry consensus that the flexible source and load of the distribution network participate in the regulation in the form of clusters. Relying on the “centralized-distributed” regulation architecture including “cluster partition

external feature calculation

instruction decomposition

and disaggregation”

we constructed a flexible re-source partitioning index system and implementation strategy after considering both the scheduling performance of the distribution network and the profit of the aggregator. Firstly

we designed a comprehensive evaluation index system to measure the performance of different cluster division methods participating in power grid dispatching and the income level of aggregation subjects. Then

aiming at the sample shortage of cluster division method that might be encountered in the actual execution process

we designed a sample generation and screening mechanism after considering the active and reactive power-voltage coupling characteristics of the distribution network. At the same time

in the process of “external feature calculation”

we proposed a dimension reduction model simplification method of flexible resource output range

which effectively reduced the difficulty of solving external feature of clusters. In the process of “disaggregation”

we introduced the economic profit objective function of adjustable objects

and the fuzzy method was used to transform this objective and the minimum scheduling error into a single objective function

which took into account the scheduling performance and the economic profit of adjustable resources in the cluster partition process. The simulation analysis results show that the excellent cluster partitioning scheme obtained by the proposed method not only has the instruction tracking performance comparable to independent control

but also increases the economic benefit by 6.4%

and the active network loss and voltage deviation rate are reduced by 1.2% and 0.6%

respectively.