The representative day planning method can improve the efficiency of integrated energy system station-network planning

but most of the existing representative day selection methods are based on data feature

ignoring the nonlinear characteristics of the planning model itself

while taking into account the existence of nonlinear feature in the selection of representative days can make the selected representative days more accurate. For this purpose

a representative day selection method based on planning decision feature is proposed

where the clustering object is converted from forecast data feature to planning decision feature

and representative days containing nonlinear information of the planning model are obtained. The preliminary planning decision is used to execute the operation problem on each forecast day

the extreme day is selected based on the loss of load volume

and different weights are assigned to representative and extreme days in the objective function. The station-network planning model of the electric-thermal integrated energy system is established with the lowest operating cost

and different numbers of representative days and extreme days are selected for planning in different ways. The results of the example show that the representative day selection based on the feature of the planning decision is able to obtain more accurate planning decisions by using only a small number of representative days

and at the same time

the load loss of the planning decision considering the extreme days is significantly reduced

which verifies the validity of the proposed selection method.