Distributed multi-objective African vulture accelerated optimization intelligent algorithm for multi-objective economic dispatch of power systems

Abstract

The context of expanding power system scale and rapid development of the power market brings many challenges to economic dispatch. Although distributed multi-objective optimization methods are more convenient in solving large-scale systems, distributed multi-objective optimization methods still have shortcomings such as inefficient economic dispatch of the system and long computation time. This study combines the acceleration of deep neural networks with distributed multi-objective optimization, constructs a novel FlattenSelf-attentionNet structure, and proposes a distributed multi-objective African vulture accelerated optimization algorithm (DMOAVAOA) to enhance computational efficiency and reduce the dispatch time of the whole system. Experimental results in the American midwestern 118-bus and the 1790-bus system indicate that: Qu et al. (2018) (1) in the American midwestern 118-bus system, compared with distributed optimization, distributed accelerated optimization reduced carbon emissions by 9.95 %, cost by 2.16 %, and total operating time by 12.38 %; Qu et al. (2019) (2) in a 1790-bus system, the distributed accelerated optimization reduced carbon emissions by 3.5 %, cost spend by 4.8 %, and total system operating time by 43.25 % compared to distributed optimization; Chen et al. (2019) (3) the DMOAVAOA proposed in this study outperforms the distributed optimization method in the evaluation of performance metrics.