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
