Application of CEEMDAN Algorithm in MPPT of Grid Photovoltaic System

In the context of the global energy crisis photovoltaic power(PVP) generation, as a clean energy source, occupies a large proportion of the total energy structure. This paper proposes to use the DBN-XGBoost hybrid algorithm based on PV sequence decomposition and reconstruction to decompose the existing PV system data, to obtain a pattern function sequence with stronger time-domain characteristics, so as to realize the extraction of the time-scale characteristics of the PV system. On this basis, two algorithms, Deep Belief Network (DBN) and Extreme Gradient Boosting (XGBoost), are used to realize the modal composition prediction of the structural model. In terms of orthogonal performance, it can be seen from the data that the orthogonality result of the EMD algorithm is 0.09523, and the orthogonality results of EEMD and CEEMDAN are only 0.01687 and 0.00548. Finally, the simulation results show that compared with other PVP prediction models, the CEEMDAN-DBN-XGBoost PVP prediction model has higher prediction accuracy.