Photovoltaic Region Prediction Based on Improved Convolutional Neural Network and Cluster Analysis

At present, the photovoltaic cluster power generation form integrated by multiple photovoltaic power plant is becoming more and more common. The accuracy of predicting the output power of photovoltaic clusters is very important for the safe and economical operation of power systems. In order to further improve the accuracy of PV cluster power prediction, a new cluster power prediction method based on Fuzzy C-Means cluster analysis theory considering weather classification and improved convolutional neural network is proposed in this paper. Firstly, photovoltaic power plants in a certain region are classified based on cluster analysis. Secondly, the improved convolutional neural network uses different convolution layers and pooling layers to process the data according to the results of cluster analysis, and outputs the predicted value of photovoltaic cluster power. Finally, the data of five PV plant in a region of Australia are used to verify the effectiveness of the proposed method.