Article in Press
1. Generalized versus Panel-Specific Machine Learning Models for Photovoltaic Power Estimation under Limited Multi-Module Field Data
Author : Giovanni Dimas Prenata*, Ahmad Ridhoi
Abstract
Limited multi-module field data create a trade-off between generalized models that pool observations from several photovoltaic modules and panel-specific models that preserve module-dependent behavior but use fewer training samples. This study compares three modeling scopes for photovoltaic power estimation using 63 observations from three nominally 100 W modules measured over four days. Generalized-Shared, Generalized-Panel-Aware, and Panel-Specific scopes were evaluated using Ridge Regression, K-Nearest Neighbor Regression, Radial Basis Function Network Regression, and Random Forest Regression under nested leave-one-day-out validation. A datasheet-informed physical power estimate was also evaluated as a non-machine-learning baseline. Panel-Specific Ridge achieved the best overall performance with an RMSE of 7.3236 W, MAE of 5.6750 W, and R2 of 0.8788, reducing RMSE by 60.23% relative to the physical baseline (18.4161 W). Generalized-Panel-Aware Random Forest achieved an RMSE of 7.7876 W overall and became the best configuration for P3 (6.3310 W) and for the Tuesday left-out day (4.9565 W). Generalized KNN and panel-aware RBF also benefited from pooled training data, whereas Panel-Specific Ridge remained the strongest regularized linear configuration. Wilcoxon signed-rank tests with Holm correction found no statistically significant scope differences across the 12 panel-day units; the largest scope effect was observed for panel-aware Random Forest (rank-biserial = -0.4872, adjusted p = 0.3027). The results show that the benefit of pooling or specialization is algorithm- and operating-condition-dependent and should be interpreted within the limitations of the small single-site field dataset.
Accepted Date : 12 September 2026
2. Quality Prediction Based on Artificial Neural Networks in the Plastic Injection Molding Process
Author : Riana Magdalena Silitonga*, Ferdian Aditya Pratama , Stefani Prima Dias Kristiana
Abstract
This study proposes a data-driven approach for predicting product quality in plastic injection molding based on actual manufacturing data. The dataset comprises 500 production observations, including 461 non-defective and 39 defective products. The data were divided into training and testing sets using an 80:20 ratio, with SMOTE applied exclusively to the training set to mitigate the substantial class imbalance. Regression analysis was subsequently conducted to identify the most relevant process parameters from five candidate variables. This analysis identified melt end point and melting time as statistically significant predictors, which were then used to develop three classification models: Backpropagation Neural Network (BPNN), Random Forest, and Support Vector Machine. Each model was evaluated over ten experimental repetitions, with Accuracy and AUC serving as the primary performance measures, while Precision, Recall, F1-score, and confusion matrices provided additional evaluation. Across the experiments, BPNN produced the strongest overall predictive performance. The main contribution of this study is the integration of statistically informed feature selection and imbalance treatment with computationally lightweight machine-learning models, demonstrating a practical approach to quality prediction when only a limited amount of real-world manufacturing data is available.
Accepted Date : 12 September 2026
3. Analysis of Power Cable Sizing and Voltage Drop in the Upgrading Stasiun Pengumpul Puspa Asri 2 Project
Author : Putri Wulandari, Linda Wijayanti*
Abstract
The upgrading of oil and gas production facilities requires an electrical power distribution system that is reliable, safe, and compliant with voltage quality limits. This article discusses power cable sizing and voltage drop evaluation for the Puspa Asri Gathering Station Phase 2 Upgrading Project in Jambi, based on engineering documents for calculating cable sizing and voltage drop in the Puspa Asri Phase 2 Collector Station Upgrading Project. The method includes load identification, full-load current calculation, determination of cable current-carrying capacity after temperature and grouping correction factors, verification of voltage drop under steady-state and motor-starting conditions, and assessment of thermal withstand capability under short-circuit conditions. The design criteria refer to IEC 60364-5-52 and related standards, with voltage drop limits of 2% for switchgear/MCC-to-switchgear/MCC routes, 5% for motor terminals under normal operating conditions, 15% during motor starting, and 3% for distribution panel feeders. The evaluation of 16 main cable routes shows that all selected cables satisfy the requirements for ampacity, voltage drop, and short-circuit withstand capability. The selected cable sizes range from 3C+E 4 mm² to 4C+E 95 mm², using a temperature correction factor of 0.91 and a grouping derating factor of 0.79 for aboveground installation. The Water Injection Pump C case study shows that the selected 4C+E 35 mm² cable has a corrected current-carrying capacity of 90.58 A, which is greater than the design current of 47.03 A, with a steady-state voltage drop of 1.33% and a motor-starting voltage drop of 3.95%. Therefore, the cable design is considered technically feasible to support reliable operation of the production facility.
Accepted Date : 31 Juli 2026
4. An mRMR-Based Feature Optimization Framework for Gaming Addiction Classification
Author : Rizki Surya Permana*, Arif Rahman Hakim, Seshariana Rahma Melati, Muhammad Farhan Maulana
Abstract
This Gaming-addiction classification is challenging because problematic gaming behavior is influenced by multiple behavioral, psychological, lifestyle, and contextual factors, while redundant or weakly informative features may reduce the effectiveness of machine-learning models. To address this issue, this study investigates minimum Redundancy Maximum Relevance (mRMR) feature selection to obtain a more compact and discriminative feature representation for gaming-addiction classification. Random Forest, Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), and K-Nearest Neighbors (KNN) were evaluated using two configurations: a baseline using the complete predictor set and an mRMR-based configuration using the 15 highest-ranked features. All preprocessing, feature selection, hyperparameter tuning, and cross-validation were performed on the training data, while the independent test set was used only for final evaluation. The results show that the effect of mRMR varies across classifiers, indicating that feature reduction does not universally improve classification performance. The strongest improvement was obtained by SVM, whose accuracy increased from 0.940 to 0.960 and F1-score from 0.824 to 0.857 after applying mRMR. The mRMR–SVM model also achieved a precision of 1.000 and a recall of 0.750. In contrast, the baseline XGBoost model achieved a higher recall of 0.914, indicating a trade-off between false-positive reduction and positive-case detection. These findings demonstrate that mRMR can improve classification when the selected feature space is well aligned with the learning mechanism of the classifier, particularly for margin-based SVM. The selected features further indicate that gaming-addiction classification is associated not only with gaming intensity but also with psychosocial and lifestyle characteristics. Therefore, the proposed mRMR–SVM configuration provides a compact and effective representation for gaming-addiction classification, although validation on larger and more diverse datasets is required before broader generalization.
Accepted Date : 12 September 2026
