Photovoltaic Power Regression Model Based on Gauss Boltzmann Machine

Improving the short-term power forecast of photovoltaic panels is a key issue for solar photovoltaic power generation to effectively merge from distributed power sources into the current large-scale power grid, and has great significance for improving the utilization of solar photovoltaic power generation. The restricted Boltzmann machine is an autoencoder that can be used to build deep learning models, which can reconstruct the input data. In this paper, a restricted Boltzmann machine is introduced on the basis of a linear regression model, and a predictive regression model for photovoltaic power generation is constructed. The model first reconstructs the original data using a restricted Boltzmann machine, then builds a linear regression model on the reconstructed data, and finally applies it to the photovoltaic panel output power data provided by GEFCom2014. Experiments show that the restricted Boltzmann machine can greatly enhance the regression prediction ability of general regression models.