Dual-stage attention-based long-short-term memory neural networks for energy demand prediction
Dual-stage attention-based long-short-term memory neural networks for energy demand prediction
复制标题
用于能源需求预测的双阶段基于注意力的长短期记忆神经网络
DOI：
10.1016/j.enbuild.2021.111211
复制
发表时间：
2021
影响因子：
6.7
通讯作者：
Ovtcharova Jivka
中科院分区：
文献类型：
--
作者：
Peng Jieyang;Kimmig Andreas;Wang Jiahai;Liu Xiufeng;Niu Zhibin;Ovtcharova Jivka
住宅建筑的能源需求预测在智慧城市的运营中发挥着重要作用，因为它构成了城市能源系统规划和运营决策的基础。深度学习算法通常用于可靠地预测潜在的能源使用，因为它们可以克服相对于标准回归模型的能源预测中对长距离数据的依赖性问题。然而，在能源预测方面仍然有两个问题需要解决，即分类特征的编码和用于进行预测的最相关特征的自适应提取。针对上述问题，提出了一种基于嵌入机制和两阶段注意力长期记忆神经网络的中长期能源需求序贯预测模型。最后，以上海浦东地区居民住宅三年的日用电量数据为例，对模型的有效性进行了实证研究。预测结果表明，该模型可以有效地提取关键特征，是高度相关的能源消费动态，利用长期依赖的时间序列数据。此外，混合模型在长期预测方面比其他模型显示出更好的稳定性。这项工作还讨论了深度学习在能源领域应用的未来研究机会和可能性。
Forecasting energy demand of residential buildings plays an important role in the operation of smart cities, as it forms the basis for decision-making in the planning and operation of urban energy systems. Deep learning algorithms are commonly used to reliably predict potential energy usage since they can overcome the issue of dependency on long-distance data in energy forecasting relative to the standard regression model. However, there are still two problems to be solved in terms of energy forecasting, the coding of categorical characteristics and adaptive extraction of the most relevant characteristics for use in making predictions. To solve the above problems, we proposed a sequential forecasting model for medium- and long-term energy demand forecasting, based on an embedding mechanism and a two-stage attention-based long-term memory neural network. Then an empirical study was conducted on three years of daily power consumption data from residential buildings in the Pudong district of Shanghai to verify the validity of the model. The forecasting results show that the model can effectively extract key features that are highly correlated with energy consumption dynamics by employing long-term dependencies in the time series data. Additionally, the hybrid model shows improved stability over other models in terms of long-term forecasting. This work also discusses future research opportunities and possibilities for deep learning applications in the energy sector.