计算机技术与发展2026，Vol.36Issue(1)：147-155,9.DOI:10.20165/j.cnki.ISSN1673-629X.2025.0181
基于特征融合和DTCN-BiLSTM-TSA的光伏功率预测
Photovoltaic Power Prediction Based on Feature Fusion and DTCN-BiLSTM-TSA
摘要
Abstract
Photovoltaic power is easily affected by the randomness and volatility of meteorological and light conditions.In order to improve the prediction accuracy of photovoltaic power generation,we propose a combined model based on DTCN-BiLSTM-TSA.Firstly,Pearson correlation coefficient is used to analyze the correlation between photovoltaic power generation and meteorological and light conditions,and the power of multi-feature fusion is used to form a multivariate time series as input.Then,the dual time convolution network(DTCN)is used to capture the global and local features in the time series data,and the bi-directional long-term and short-term memory network(BiLSTM)is used to further obtain the long-term and short-term dependencies.Finally,the temporal-spatial attention(TSA)mechanism obtains key features and spatio-temporal correlations from the output matrix of BiLSTM to improve the robustness of the model.The experimental results show that there is a strong correlation between photovoltaic power generation and external factors.We use key indicators such as RMSE,MAPE,MAE and R2 to evaluate the performance of the model.The DTCN-BiLSTM-TSA model is superior to other deep learning models in terms of prediction accuracy and robustness.关键词
光伏发电/功率预测/时间卷积网络/双向长短期记忆网络/时空注意力Key words
photovoltaic power generation/power prediction/temporal convolutional network/bi-directional long short-term memory/temporal-spatial attention分类
信息技术与安全科学引用本文复制引用
马瑞,朱东歌,帅春燕,刘佳,沙江波,康文妮..基于特征融合和DTCN-BiLSTM-TSA的光伏功率预测[J].计算机技术与发展,2026,36(1):147-155,9.基金项目
宁夏自然科学基金项目(2023AAC03854) （2023AAC03854）
