Accurate prediction of electric energy consumption is critical for both user convenience and supplier efficiency. This study introduces an ensemble approach that integrates clustering algorithms with machine learning (ML) models to enhance prediction accuracy by identifying consumption patterns within buildings. The research focused on residential apartments in the metropolitan area of Korea, utilizing four evaluation methods (Elbow-Method, Silhouette Score, Calinski-Harabasz Index, and Dunn Index) across five data collection intervals (10 min, 1 h, 1 day, 1 week, and 1 month). Five ML models (CatBoost, Decision Tree, LightGBM, Random Forest, XGBoost) were assessed for their prediction performance across clusters. Additionally, ML models that exhibited high performance within each cluster were amalgamated into an ensemble model to assess the predictive performance regarding total electric energy consumption at the research site. Optimal clustering resulted in two clusters (142 houses for C0, 206 houses for C1) using monthly resampled power data. CatBoost and LightGBM exhibited the highest average prediction performance. Based on the possible combinations of the two models applied to each cluster, four ensemble models were developed: CB-CB, CB-LGBM, LGBM-CB, and LGBM-LGBM. Statistical analysis confirmed that all ensemble models significantly outperformed the control group's traditional ML approaches without clustering (p < 0.05 or 0.01). The proposed clustering-based ML ensemble model in this study can predict the energy consumed in buildings more accurately by accounting for the unique consumption pattern of each house. It is anticipated to contribute effectively to energy consumption reduction.

District heating (DH) networks are a key component of low-carbon urban heating in the future, as greenhouse gas emissions and sustainability concerns drive the heating sector to transform itself. DH is not a new technology, but it has been constantly evolving. The latest generation of DH facilitates the distribution of low-temperature renewable heat sources. In recent years, most studies have focused on managing peak demand, improving low-carbon technologies, and improving load prediction. However, there is a risk of misinterpretation, as recent generations of DH, which operate at significantly lower temperatures than conventional DH, are being developed simultaneously. This review aims to analyze the different definitions of the fifth-generation district heating and cooling (5GDHC) and introduce a straightforward concept of this new technology. It also describes the potential strengths, weaknesses, and challenges of integrating 5GDHC into existing systems, as well as practical recommendations. Finally, it analyzes the crucial components and notable characteristics of 5GDHC to provide a clear picture of its evolution and uniqueness.

High integration of semiconductor processes is being made to realize high performance in miniaturized chips. The performance of a semiconductor chip may vary depending on target variables such as thickness, line width, shape, composition, and physical properties of each layer constituting the chip. Therefore, in order to secure chip performance, accurate detection of target variable values and quality control are required, and it is necessary to check in advance for defects that may occur during the process. Optical inspection technology is widely used in the semiconductor metrology field due to its advantage in that it can detect defects in the wafer at high speed by scanning the wafer with a light source having a specific wavelength band. However, in recent years, the size of defects caused by high integration and miniaturization of semiconductor chip processes is getting smaller, and thus there is a limit to detecting micro defects using conventional optical methods. In this study, we propose an algorithm to improve the defect detection performance by utilizing multi-scan images acquired under various conditions. Using the suggested algorithm, it was confirmed that the SNR (Signal to noise ratio) of the defect of interest was improved by about 99%, and the classification performance for noise was improved by 4 times.

Heat usage patterns, which are greatly affected by the users' behaviors, network performances, and control logic, are a crucial indicator of the effective and efficient management of district heating networks. The variations in the heat load can be daily or seasonal. The daily variations are primarily influenced by the customers' social behaviors, whereas the seasonal variations are mainly caused by the large temperature differences between the seasons over the year. Irregular heat load patterns can significantly raise costs due to pricey peak fuels and increased peak heat load capacities. The in-depth analyses of heat load profiles are regrettably quite rare and small-scale up until now. Therefore, this study offers a comprehensive investigation of a district heating network operation in order to exploit the major features of the heat usage patterns and discover the big factors that affect the heat load patterns. In addition, this study also provides detailed explanations of the features that can be considered the main drivers of the users' heat load demand. Finally, two primary daily heat usage patterns are extracted, which are exploited to efficiently train the prediction model.

This research proposes a methodology for the selection of input variables based on eXplainable AI (XAI) for energy consumption prediction. For this purpose, the energy consumption prediction model (R2 = 0.871; MAE = 2.176; MSE = 9.870) was selected by collecting the energy data used in the building of a university in Seoul, Republic of Korea. Applying XAI to the results from the prediction model, input variables were divided into three groups by the expectation of the ranking-score (Fqvar) (10 ≤ Strong, 5 ≤ Ambiguous < 10, and Weak < 5), according to their influence. As a result, the models considering the input variables of the Strong + Ambiguous group (R2 = 0.917; MAE = 1.859; MSE = 6.639) or the Strong group (R2 = 0.916; MAE = 1.816; MSE = 6.663) showed higher prediction results than other cases (p < 0.05 or 0.01). There were no statistically significant results between the Strong group and the Strong + Ambiguous group (R2: p = 0.408; MAE: p = 0.488; MSE: p = 0.478). This means that when considering the input variables of the Strong group (Fqvar: Year = 14.8; E-Diff = 12.8; Hour = 11.0; Temp = 11.0; Surface-Temp = 10.4) determined by the XAI-based methodology, the energy consumption prediction model showed excellent performance. Therefore, the methodology proposed in this study is expected to determine a model that can accurately and efficiently predict energy consumption.

This study proposed a deep-learning-based approach that generates synthetic high-resolution log data from original-resolution log data for accurate reservoir characterization, where the resolution of the synthetic data is comparable to that of core data. The reliability of the proposed approach was tested with application to the Volve oil field in Norway using three deep-learning algorithms (i.e., deep neural network, convolutional neural network, and long short-term memory). These deep-learning algorithms were employed to generate high-resolution sonic log data from other log-type data. The overall performance of each algorithm was acceptable. In particular, the long short-term memory algorithm yields a coefficient of determination greater than 0.9 when the high-to-original-resolution ratios are two, five, and ten. We anticipate that the proposed model can be used to derive logging-based reservoir parameters with a resolution that is comparable to that of core-based reservoir parameters.