今日获知，本科生郭继煊的论文Scaling-Risk Early Warning in Industrial Cooling-Water Systems: A Chemically Regularized Machine-Learning Framework Using Routine Water-Quality Data被Journal of Water Process Engineering接收。论文得到国家自然科学基金、陕西省杰出青年科学基金的资助。相关数据得到山西飞宇环境工程股份有限公司和欣格瑞（山东）环境科技有限公司的有力支撑。摘要如下：

Scaling in industrial recirculating cooling-water systems reduces heat-transfer efficiency, increases energy consumption, and raises maintenance costs, while direct measurements of deposit thickness, scaling rate, or heat-transfer resistance are seldom available in routine operation. This study proposes a machine-learning framework for the forecasting of a surrogate scaling-risk indicator derived from routine water-quality monitoring data. A hybrid risk-indicator target is constructed by integrating a chemically interpretable Puckorius Scaling Index (PSI)-derived component with a principal component analysis (PCA)-derived component that captures facility-specific water-quality variation. A chemically regularized Transformer (CR-Transformer) then learns the temporal relationship between the previous 10 monitoring steps and the next risk-indicator score. The framework was evaluated using three real industrial datasets and achieved high predictive accuracy, with single-scenario R2 values of 0.9870, 0.9849, and 0.9750. Weight-sensitivity analysis suggested that the indicator remained stable over a moderate range of fusion weights, while PCA correlation and SHAP analyses supported the chemical interpretability of both the indicator and the predictor. These results indicate that the proposed framework can provide an early-warning reference for water-quality conditions prone to calcium carbonate (CaCO3) scaling.