Grid-connected equivalent modeling of DC microgrid based on optimized extreme learning machine

Aiming at the grid-connected modeling problem of DC microgrid, a grid-connected equivalent mode-ling method of DC microgrid based on optimized extreme learning machine is proposed. The voltage and power data of the grid-connected point of DC microgrid are taken as the input and output of the extreme learning machine respectively, the grid-connected equivalent model of DC microgrid based on extreme lear-ning machine is constructed. In the initialization process of extreme learning machine, input weight and hidden layer node bias are randomly set without any change later, which leads to the lack of adaptability in modeling and affects the modeling accuracy. The shark smell optimization algorithm is used to optimize the input weight and hidden layer node bias of the extreme learning machine, so as to improve the mode-ling accuracy. Shark smell optimization algorithm is a highly efficient optimization algorithm, which can simu-lates the hunting process of sharks for optimization, and the concentration of smell particles guides the updating of shark position. By comparing with the actual simulation model of microgrid, the rationality and accuracy of the modeling method are verified, indicating that the model has good practical application value. © 2020, Electric Power Automation Equipment Press. All right reserved.