Photovoltaic Power Generation Forecasting Based on Attention-CNN-BiLSTM Model
Feipeng Huang, Zhen Pan, Zeyi Huang, Zhaoyu Li, Lidan Chen, Langwen Zhang · 2024
With the rapid advancement of computer technology, the processing and storage capacity of computers have significantly improved. This has led to the remarkable development of artificial intelligence and deep learning. Among these advancements, neural network models such as Long Short-Term Memory (LSTM) have shown impressive results in AI prediction. LSTM addresses the issues of gradient vanishing and explosion, providing a theoretical basis for deep learning. In this work, a photovoltaic power generation forecast model is investigated based on Convolutional neural networks (CNN)-bi-directional LSTM (BiLSTM) network with attention mechanism. Convolutional neural networks is we incorporated with BiLSTM network, constructing a CNN-BiLSTM network. An attention mechanism is introduced for CNN-BiLSTM to improve the photovoltaic power generation forecasting accuracy. Our proposed algorithm demonstrates superior prediction accuracy compared to traditional LSTM model.