On Predicted Research Methods of Supply Capacity of Micro-grid Based on Improved Particle Swarm Optimization

Micro-grid is a kind of distributed low-voltage supply network by integrating various distributed powers, energy storage systems and controlled load. In micro-grid, distributed power can be divided into certain and stochastic power. The prediction on supply capacity of micro-grid focuses on electric energy production by stochastic power. The wind energy power is predicted in this paper by support vector machine (SVM) and the combination method of improved particle swarm optimization (PSO) and simulated annealing (SA) which forms a hybrid algorithm of SA-IPSO to optimize SVM model parameter adaptively. The case study have proved that this algorithm adjusts model parameter by adaptive learning, so this predicted model track the fluctuation and change of wind energy power effectively and further predict the total supply capacity of micro-grid more accurately.