Miao Wang

In enclosed cold-aisle data-center rooms, hotspot temperatures within the cold aisles are directly related to server inlet air safety and air-conditioning operation performance. Therefore, high-accuracy temperature prediction is of great significance for thermal-risk warning and operational optimization. This study focuses on an enclosed cold-aisle data-center room in a large data center located in a hot-summer and warm-winter region. The maximum temperature of one selected cold aisle is taken as the prediction target, and a hybrid prediction model integrating TSception, Transformer, and TCN is proposed to characterize multi-scale local disturbances and cross-period temporal dependencies. The model inputs and historical time window are determined through cold–hot separation analysis, cold-aisle temperature-stability comparison, lag analysis between supply air temperature and cold-aisle temperature, and correlation analysis of candidate variables. The experimental results show that the proposed model achieves an MAE of 0.285 °C, an RMSE of 0.438 °C, an NRMSE of 4.76%, and a MAPE of 1.13% in predicting the cold-aisle temperature of the AB aisle, outperforming Persistence, BP, LSTM, XGBoost, and several ablation models. Further multi-aisle experiments demonstrate that the proposed model consistently maintained low prediction errors in independent prediction tasks across different cold aisles, indicating good applicability of the model architecture for characterizing temperature time series in multiple cold aisles within the same data-center room. The proposed method can provide a reference for hotspot-risk identification in enclosed cold aisles and operational optimization of precision air-conditioning systems.

Accurate electricity carbon emission factors are crucial for assessing overall social carbon emissions and achieving China’s “dual carbon” goals. This paper proposes a dynamic correction model that integrates lifecycle extension, power exchange networks, and multi-time-scale decomposition to address the limitations of static carbon emission factors. The model considers factors such as power generation structure, cross-regional transmission, clean energy proportion, line losses, and non-CO₂ greenhouse gas emissions, and achieves dynamic correction at quarterly and monthly scales, enhancing timeliness and regional adaptability. Results show that transmission losses, energy structure, and inter-provincial electricity exchange significantly impact carbon emission factors. For instance, in 2022, line losses in Xinjiang and Gansu raised the electricity carbon emission factor by over 0.06 kgCO₂/kWh. Monthly factors exhibit significant seasonal fluctuations, with some regions showing variations of up to 105% of the annual average. Areas rich in hydropower, such as Yunnan, Sichuan, and Qinghai, experience pronounced fluctuations, highly sensitive to changes in water volume, offering more accurate reflections of carbon emission changes during electricity consumption. This study presents a refined dynamic correction method for electricity carbon emission accounting, providing theoretical support for carbon emission policy development and performance evaluation.

Energy consumption forecasting offers a foundation for governmental agencies to establish energy consumption benchmarks for public institutions. Meanwhile, correlation analysis of institutional energy use provides clear guidance for building energy-efficient retrofits. This study employed five machine learning models to train and predict monthly energy consumption intensity data from 2020 to 2022 for three types of public institutions in China’s eastern coastal regions. A novel ensemble model was proposed and applied for energy consumption prediction. Additionally, the SHAP model was utilized to analyze the correlation between influencing factors and energy consumption data. Finally, the relationship between climatic factors and monthly energy consumption intensity was investigated. Results indicate that the ensemble model achieves higher predictive accuracy compared to other models, with regression metrics on the training set generally exceeding 0.9. Although XGBoost also demonstrated strong performance, it was less stable than the ensemble model. Energy intensity across different building types exhibited strong correlations with the number of energy users, floor area, electricity use, and water consumption. Linear analysis of temperature and energy consumption intensity revealed a directional linear relationship between the two for both medical and administrative buildings.