Optimal energy management of grid-connected photovoltaic micro-grid

Abstract

In accordance with the energy optimization management of grid-connected PhotoVoltaic (PV) micro-grid, this paper firstly forecasts the output of PV power based on Least Squares Support Vector Machine (LS-SVM) to resolve the randomness of distributed PV power generation. Then a Micro-grid Energy Management (MEM) model that treats the lowest total cost of electricity in micro-grid as the objective function is set up, which takes the power balance, the upper and lower power of units, the unit commitment status, PV as well as the grid real-time bidding pricing and so on into account. Next a Modified Artificial Fish School (MAFSA) algorithm, which adds dynamic parameters (changeable visual distance, variable moving step) and Student's t-distribution probability operator to enhance the global optimization ability and the convergence accuracy in the mid-to-late iteration, is introduced to economically allocate all the micro powers in micro-grid. Finally, through an example analysis, the results show that the proposed MEM model is feasible and has a certain reference value for MEM.