Analysis of Power Load Components Based on Neural Network

Dynamic load model is of great value in analysis and simulation of power grid, while the composition of load has an important influence on the characteristic and accuracy of load model. There is mass of load equipment in power grid, so it is usually necessary to classify the load equipment according to their different characteristic first, then establish a suitable load model separately, and synthesize them into a composite model according to their proportion. Only the power value of load can be measured when power grid operates in a normal and steady state, and their dynamic response characteristics are unknown. Therefore, it is difficult to obtain sufficient information for load classification. However, frequency characteristics of load are distinguished and easy to obtain, so it is useful for load classification indirectly. Firstly, several elector-magnetism models of typical load are established, and their harmonic curves and frequency characteristic are simulated and analyzed separately. Then, cluster analysis is carried out to obtain the typical characteristics and the membership degree which can represent the contents of load overall. Finally, a classification method combined with Neural Network is adopted, and the computation speed is also improved.