EcoPredict: A Unified Deep Learning Framework for Forecasting Energy Consumption in Cloud Classrooms
Xie Jintao, Lei Yuan · 2024
As traditional teaching methods are gradually shifting towards online education models, cloud classrooms, as a new type of teaching infrastructure, have brought energy consumption management to the forefront as a key issue. To address this challenge, this study introduces a deep learning framework named EcoPredict. EcoPredict integrates Long Short-Term Memory networks (LSTMs) and Dense Networks to effectively process and predict dynamic changes in energy consumption, thereby supporting more rational energy allocation strategies. This research utilized actual energy usage data from a university cloud classroom in China for its experiments. The results demonstrate that the EcoPredict framework significantly outperforms traditional prediction models, achieving an accuracy rate of 93.74%, which is considerably higher than the performance of using either Dense Network or LSTM models alone. Through in-depth analysis of the EcoPredict framework, this study not only improves the accuracy of predicting energy consumption in cloud classrooms but also provides effective technical support for the rational allocation of online education resources.