A New Prediction Approach Using FCM and Deep Cascade Elastic Broad Learning for Photovoltaic Power Generation in Short Term

Abstract

In recent years, photovoltaic power generation prediction has mainly faced some difficulties, such as low prediction accuracy, time-consuming model training, and so on. In order to improve the prediction accuracy of photovoltaic power generation in short term, it is greatly beneficial to optimize power allocating plans and improve economic benefits. Therefore, a new prediction method of Photovoltaic (PV) power generation based on Fuzzy C-Means (FCM) and deep cascade elastic broad learning is proposed in this paper. Firstly, the fuzzy c-means clustering algorithm is used to classify the input factors. Then, the feature mapping nodes of the broad learning model are constructed by the deep cascade method to extract the features of the input factors fully. The elastic net regularization method is employed to constrain the network output weight, measure the influence of the nodes on the weight, extract the more influential nodes, and sparse network structure. A deep cascade elastic broad learning network is established for short-term photovoltaic power prediction. Finally, the effectiveness of the method is verified through the original data. The experimental results verify the model can better predict the actual photovoltaic power generation and has a good application value.