Implementing photovoltaic-energy storage-direct current-flexible(PEDF) technology to build green buildings is crucial. Fully tapping into the energy-saving potential on the demand side of buildings is also essential. Both efforts are vital for achieving sustainable energy development and building a green

low-carbon society. However

the large and heterogeneous load groups in buildings pose significant challenges to developing unified regulation methods and standards. This paper proposes an energy management strategy for PEDF buildings that considers the multiple comfort levels of heterogeneous load clusters to these address challenges. First

to address the difficulty of unified modeling for a large number of heterogeneous loads

this paper presents an improved K-means algorithm based on dynamic time warping(DTW) and secondary clustering of load operation periods. Models are then established based on the clustering results

and the effectiveness of the clustering method is verified using typical daily load data from summer and winter seasons. Then

considering the multi-parameter coupling effects of the indoor environment

an economic-multi-comfort coupling model is constructed. This model integrates economic factors with multidimensional comfort indicators

including thermal comfort

air quality comfort

auditory comfort

and visual comfort. Furthermore

this paper employs a non-dominated sorting genetic algorithm II(NSGA-II) algorithm to conduct multi-objective optimization of typical flexible loads

as well as aggregated equivalent energy storage from building-owned energy storage and electric vehicle aggregated equivalent energy storage(EV-AES). Finally

simulation verification of the load within a PEDF building is performed using a typical day in summer as examples.