基于注意力机制的CNN-LSTM建筑能耗预测方法研究
Research on a CNN-LSTM building energy consumption prediction method based on attention mechanism
摘要：
建筑能耗分析预测是提高建筑用能效率的关键技术,是响应国家“双碳”战略的重要手段。由于建筑能耗数据具有强时序性特点,利用传统的深度学习技术难以有效提取数据中的高维特征,且易丢失重要信息。为此,本文提出了一种基于注意力机制的CNN-LSTM建筑能耗预测方法,该方法利用CNN提取能耗数据中的空间特征、LSTM处理时序数据、注意力机制确定特征权重,提高了模型预测精度。
Abstract：
The analysis and prediction of building energy consumption is a key technology to improve the energy efficiency of buildings and an important means to address the national “dual-carbon” strategy. Due to the strong temporal characteristics of building energy consumption data, it is difficult to effectively extract high-dimensional features in the data by using traditional deep learning techniques, and it is easy to lose important information. Therefore, this paper proposes a CNN-LSTM building energy consumption prediction method based on the attention mechanism, which uses CNN to extract spatial features in energy consumption data, LSTM to process time series data, and attention mechanism to determine feature weights, improving the prediction accuracy of the model.
Keywords:building energy consumption; prediction; deep learning; convolutional neural network; long short-term memory network; attention mechanism