Article Per Year (5 Year)
p-Index From 2020 - 2025
0.61
P-Index
This Author published in this journals
All Journal Bulletin of Electrical Engineering and Informatics Buletin Ilmiah Sarjana Teknik Elektro
Nguyen, Trung Dung
Unknown Affiliation
Electrical & Electronics Engineering

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Buletin Ilmiah Sarjana Teknik Elektro

 1 
Analysis of Swarm Size and Iteration Count in Particle Swarm Optimization for Convolutional Neural Network Hyperparameter Optimization in Short-Term Load Forecasting Nguyen, Tuan Anh; Nguyen, Trung Dung
Buletin Ilmiah Sarjana Teknik Elektro Vol. 7 No. 3 (2025): September
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/biste.v7i3.13953

Abstract

Short-term load forecasting (STLF) is critical in modern power system planning and operation. However, the effectiveness of deep learning models such as Convolutional Neural Networks (CNNs) depends on selecting hyperparameters, which are traditionally tuned through time-consuming trial-and-error processes. The research contribution of this study is to systematically analyze how two key parameters—swarm size and iteration count—in Particle Swarm Optimization (PSO) affect the performance of CNN hyperparameter tuning for STLF. A CNN architecture with fixed convolutional depth is optimized using PSO over selected hyperparameters, including the number of filters, batch size, and training epochs. The experiments use two regional Australian electricity load datasets: New South Wales (NSW) and Queensland (QLD). A three-fold cross-validation strategy is employed, and the Mean Absolute Percentage Error (MAPE) is used as the primary evaluation metric. The results show that optimal PSO configurations vary significantly between datasets, with smaller swarm sizes and moderate iteration counts yielding favorable trade-offs between forecasting accuracy and computational cost. However, the reliance on MAPE, sensitivity to near-zero values, and fixed CNN architecture impose limitations. This study provides practical guidance for selecting PSO settings in deep learning-based STLF and demonstrates that tuning PSO configurations can significantly enhance model performance while reducing computational overhead. Future work may explore adaptive or hybrid optimization methods and extend to more diverse forecasting scenarios.
Hybrid Stacking of Multilayer Perceptron, Convolutional Neural Network, and Light Gradient Boosting Machine for Short-Term Load Forecasting Nguyen, Trung Dung; Tuan, Nguyen Anh
Buletin Ilmiah Sarjana Teknik Elektro Vol. 7 No. 4 (2025): December
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/biste.v7i4.14410

Abstract

Short-term load forecasting (STLF) is essential for scheduling, dispatch, and demand-side management. Real-world load series exhibit rapid local fluctuations and calendar or exogenous influences that challenge single-model approaches. This study proposes a hybrid stacking framework combining a Multilayer Perceptron (MLP), a 1-D Convolutional Neural Network (CNN), and a Light Gradient Boosting Machine (LightGBM), integrated through a ridge-regression meta-learner. The CNN extracts local temporal patterns from sliding windows of the load series, and the MLP processes tabular features such as lags, rolling statistics, and calendar/holiday indicators. At the same time, LightGBM captures nonlinear interactions in the same feature space. Base learners are trained using a rolling TimeSeriesSplit to avoid temporal leakage, and their out-of-fold predictions are used as inputs for the meta-learner. Early stopping regularizes the neural models. Experimental backtests on Queensland electricity demand data (89,136 half-hourly samples) demonstrate that the stacked model achieves markedly lower forecasting errors, with MAPE ≈ 0.81%, corresponding to a 24% reduction compared to CNN (MAPE ≈ 1.07%) and a 32% reduction compared to MLP (MAPE ≈ 1.19%). Regarding runtime, LightGBM is the fastest (25s) but least accurate, while the stacked model requires longer computation (2488s) yet delivers the most reliable forecasts. Overall, the proposed framework balances accuracy and robustness, and it is modular, reproducible, and extensible to additional exogenous inputs or base learners.
Co-Authors Ha, Chi Kien Le, Long Ho Ngo, Thanh Quyen Nguyen Anh Tuan Nguyen, Hoanh Nguyen, Tuan Anh Pham, Trung Kien