IEEE i-COSTE 2026 - NPS Australia Submission System

Beyond Two Modalities: Power–Network–Host Fusion for Electric Vehicle Charging Station Attack Detection — A Controlled Ablation Study

The first documented modality-by-attack-class coverage table for CICEVSE2024, revealing that only three of five attack classes are visible across all sensors and that Backdoor detection is a power-exclusive, unbridgeable ceiling.

The first fusion evaluation combining all three modalities, anchored against the best single modality under multi-seed paired testing, showing no statistically significant improvement over the best individual sensor for either the three-class or four-class subtasks.

A rigorously null result accompanied by the identification and correction of two methodological hazards (network session leakage and host timestamp leak), plus an honest explainability audit that explicitly treats degenerate permutation importance under a performance ceiling.

Sustainable Urban Electricity Transition in Malaysia: Feasible Renewable Energy Models and Policy Analysis

This paper’s key contribution is the first integrated PVsyst–HOMER Pro dual-simulation assessment of a Solar ATAP-compliant rooftop PV system with wind-battery hybrid for urban residential Malaysia (Kuantan, Pahang). The 6.60 kWp PV system achieves 81.62% performance ratio; the optimized PV–Wind–Battery–Grid hybrid delivers LCOE of RM0.145/kWh (73.3% below grid baseline), 18.8% IRR, and 4.96-year payback, becoming a net grid exporter under Solar ATAP’s credit mechanism. A PESTEL barrier analysis further benchmarks Malaysia’s regulatory environment (score 15/30) as substantially more enabling than Indonesia’s (23/30), directly informing NETR/NDC policy recommendations.

Weather-Aware and Perception-Driven Digital Twin for Urban Robot Navigation

Urban autonomous robots require navigation systems that can adapt to real-time environmental conditions and ensure safe path planning across complex road networks. Existing routing algorithms perform well in static settings using traditional shortest-path methods, but they often fail to account for dynamic weather conditions and integrate real-time perception, limiting their reliability during adverse scenarios when safetycritical decisions are essential. Therefore, this work proposes a digital twin framework that combines YOLOv11-based object detection trained on the KITTI dataset, live meteorological data from the Open-Meteo API, and adaptive routing applied to road networks generated via OSMnx. The system models roads as weighted directed graphs, where Dijkstra’s algorithm computes optimal paths using scenario-specific multiplicative penalties ranging from 8% for light rain to 50% for severe precipitation. An interactive IoT dashboard provides real-time monitoring of conditions and routing outcomes. Experimental validation on the urban road network of Cambridge, England (3,294 nodes, 7,227 edges) shows adaptive travel times from 12.2 minutes in clear weather to 29.1 minutes in heavy rain. Vehicle classes were detected with a strong mAP50 of 0.854 and precision of 0.917, while route lengths were adjusted from 4.33 km in clear weather to 12.77 km under severe conditions. This integration of perception, meteorology, and routing enhances the safety and robustness of autonomous urban navigation.

Voltage Regulation of Hydrogen Fuel Cell Output Using a PWM-Controlled DC-DC Boost Converter: Design, Simulation, and Experimental Validation

This paper’s key contribution is a complete, reproducible fuel cell–boost converter test platform — combining a custom EasyEDA-designed PCB, an Arduino-based 25 kHz PWM control scheme, and cross-validation across theoretical, LTspice simulation, and hardware results. It experimentally demonstrates open-loop voltage regulation of a low, load-variable PEM fuel cell output (3.3–6.8 V) up to 17.57 V, achieving 85% peak efficiency at D = 0.7. Notably, it identifies where non-idealities (R_DS(on) losses, diode drop, ESR) dominate — an efficiency plateau beyond D = 0.5 — providing a practical benchmark platform for future closed-loop and wide-bandgap converter research.

An Online Barnacles Mating Optimization-Based PI Controller for DAB Converters in EV Charging

This paper’s main contribution is a non-linear, failure-mode-based degradation rate (Rd) model for solar PV, moving beyond the standard linear/performance-ratio approach. It introduces ΔT (DEL_T) — from thermal imaging — as a novel quantitative proxy for hotspot severity, combined with I–V-derived series resistance (R_s) as key ML inputs. Two models (LSTM and FFBP) are trained and cross-validated on an independent Telangana plant, achieving low RMSE and reasonable MAPE. It also argues an economical case: ~₹57,000 sensor-based ML approach vs. ₹16–45 lakh for conventional accelerated/on-field testing.

One-Year Performance Evaluation of a 157 kWp Floating Photovoltaic System under a Tropical Monsoon Climate at UMPSA, Malaysia

The significant research contribution of this work lies in providing an empirical, long-term performance validation of a 157.20 kWp floating photovoltaic (FPV) system utilizing bifacial solar modules over a full one-year operational cycle under the specific environmental stressors of a tropical monsoon climate. While the theoretical cooling effects and high-albedo benefits of aquatic PV setups are widely discussed, real-world field data—particularly regarding the actual yield advantages and Performance Ratio (PR) of bifacial technologies on water versus land—remains scarce and highly debated. By documenting a comprehensive, full-year dataset from the system installed at the UMPSA lake, this study bridges a critical gap in literature, offering solar developers and researchers actionable insights into the seasonal performance, technical viability, and efficiency gains of scaling bifacial FPV systems in hot, humid monsoon environments.

ANN and LSTM Models for Hourly Power Prediction of a Tropical Floating Photovoltaic System

The work focuses on the application of advanced computational techniques to floating photovoltaic (FPV) systems, specifically targeting performance prediction and day-ahead power forecasting.

The significant research contribution lies in addressing the dual complexities of FPV systems—namely, their unique environmental operating conditions (such as the cooling effects of water bodies) and the inherent intermittency of solar energy. By leveraging and comparing cutting-edge machine learning and deep learning architectures, such as Deep LSTM-RNNs and Artificial Neural Networks (ANN), this research advances the state-of-the-art in predictive modeling. This contribution provides the solar energy sector with highly accurate, day-ahead forecasting frameworks that are critical for optimizing grid integration, improving operational efficiency, and validating the thermodynamic advantages of floating solar arrays over traditional land-based systems.

Techno-Economic Optimization of a Battery Energy Storage System for a Tropical Floating Photovoltaic Plant

The collective research contribution centers on the technical, economic, and operational advancement of renewable energy systems, with a particular focus on floating solar and energy storage integration. The works evaluate the techno-economic viability and carbon emission reductions of large-scale floating photovoltaic (FPV) systems, establishing their potential for regional energy roadmaps like Malaysia’s. This is complemented by broader assessments of the technical capacity of FPV systems on man-made water bodies across the United States, alongside comprehensive literature reviews detailing the overall evolution of floating solar technology.

Furthermore, the research bridges primary power generation with grid stability and modernization by examining fundamental wind and solar system operations, comparative battery storage options (such as lead-acid vs. lithium-ion) for electric mobility, and optimal sizing strategies for battery energy storage systems (BESS) to manage peak shaving under specific utility tariffs. Finally, the contribution extends into performance optimization, utilizing comparative studies on cutting-edge bifacial FPV arrays and implementing advanced machine learning techniques, such as teaching-learning-based optimization for extreme learning machines, to dramatically improve power forecasting accuracy in floating solar environments.

Sustainable Energy Management in Bangladesh: A Comparative Study with Neighboring Countries

In an energy constraints environment, role of
sustainable energy management is paramount. It ensures
optimum utilization of available resources, use of appropriate
technologies, reliable and affordable energy to consumers, and
ultimate energy justice. In this context, Bangladesh is currently
confronting with multiple cross-cutting challenges of depletion
of domestic natural gas resources, growing pressure to reduce
dependency on fossil fuels, steadily increasing reliance on energy
imports, and pressing demand for affordable electricity. This
study unfolds the importance of sustainable energy
management in Bangladesh, which can support and promote the
national dream, to becoming a middle-income country, and
fulfilling the commitment to global agenda on climate change.
The study critically analyzed the current sustainable energy
management practice in Bangladesh, and compared with the
neighboring counties. The domain area of this study is
electricity. The study adopted pragmatic worldview as research
approach, where both exploratory and explanatory methods
were applied. To conduct the study, both primary and
secondary data were used. Finally, through triangulation of
three key dimensions- energy, economy and environment, the
study identifies that Bangladesh is lagging behind from prudent
sustainable energy management practice and recommends the
necessity of a comprehensive data-driven energy policy
framework for future endeavor

Remote Sensing Imagery for Crop Yield Prediction Using Confidence-Weighted Ensemble Model with Explainable AI

1. Developed a Confidence-Weighted Ensemble model that outperformed Random Forest and XGBoost.

2. Engineered novel spectral-agronomic interaction terms, with the stress-health ratio identified as the most impactful predictor.

3. Achieved 86.46% prediction accuracy while maintaining model explainability—a critical bridge between high performance and actionable farming insights.

Conference Site: IEEE i-COSTE 2026