Current Issue [Vol. 12, No. 09] [September 2026]

The increasing demand for affordable and sustainable cooking fuels has stimulated interest in converting locally available biomass residues into briquettes. This study evaluated the physicochemical and energy characteristics of briquettes produced from sawdust (SD), rice husk (RH), and melon shell (MS) using a three-component simplex-centroid mixture design. Seven formulation points comprising single, binary, and ternary biomass combinations were evaluated at a constant biomass-to-binder ratio of 90:10, with cassava starch as binder. Briquette density ranged from 202.2 to 425.6 kg/m³, with a mean of 286.33 ± 72.87 kg/m³. Moisture content ranged from 9–12%, while volatile matter, fixed carbon, and ash contents ranged from 20–62%, 4–55%, and 12–33%, respectively. The SD/MS blend recorded the highest fixed carbon content (55%), whereas pure SD had the lowest ash content (12%). Pure RH exhibited the lowest fixed carbon (4%) and highest ash content (33%). Calorific values varied from 11.71 to 32.17 MJ/kg, with the ternary SD/RH/MS formulation recording the highest value. Boiling time for 100 mL of water ranged from 9.5–10.4 minfor the briquettes, compared with 5.6 minfor charcoal and 4.7 min for liquefied petroleum gas (LPG), indicating lower heating-rate performance under the experimental conditions. Quadratic mixture-response models yielded R² values of 0.770, 0.779, 0.562, and 0.998 for fixed carbon, moisture content, volatile matter, and ash content, respectively. The findings indicate that biomass composition influences briquette physicochemical and energy characteristics, while blending can improve selected fuel properties. The mixture design provides a basis for further formulation optimization and independent validation.
Keywords: biomass; briquettes; cassava-starch; composition; mixture; sawdust, sustainable.

The application of an AI Engine in designing a multi-channel coherent accumulator using Fast Fourier transform of phased array radar is investigated. The efficiency of signal accumulation in improving the detection of small-footprint targets in advanced phased array radars is analyzed. An overview of the AI Engine is provided. Outline the architectural design of the Versal series of components of Advanced Micro Devices, illustrating its integration with the AI Engine. The capabilities of the AI Engine and graphics processors are compared. A multi-channel signal accumulator utilizing the Fast Fourier transform in digital phased array radars, employing the Vitis Model Composer design tool, is designed. A multi-channel correlation accumulator using the Fast Fourier transform in multi-channel radar signal accumulation is simulated and compiled.
Keywords: AI Engine, Fast Fourier transform, moving target detector, digital phased array radar.

Predicting solar energy output with fine spatial and temporal resolution remains one of the most pressing challenges in distributed photovoltaic (PV) system management. This paper introduces Q-WEFS (Quantum-Enhanced IoT Weather and Energy Forecasting System), a real-time environmental intelligence platform that pairs low-cost IoT sensing hardware with a hybrid classical-quantum machine learning pipeline to enable smarter, more responsive solar energy management at the edge. An ESP32-D0WD-V3 microcontroller collects readings from four physical sensors and streams structured JSON telemetry at 5-second intervals. A RandomForestRegressor trained on 40,896 hourly NASA POWER records for Hyderabad, India (17.38°N, 78.46°E) achieves R² = 0.895 (MAE = 2.05°C) for next-day (24-hour-ahead) temperature prediction. Building on this strong classical baseline, a Variational Quantum Regressor (VQR) — using a ZZ Feature Map encoder with 5 qubits and a Real Amplitudes ansatz — learns the structured patterns left in the Random Forest’s residuals via the parameter-shift rule. The resulting hybrid system reaches R² = 0.912 (MAE = 1.82°C), a statistically significant 11.2% reduction in mean absolute error relative to the Random Forest baseline (p = 1.1 × 10⁻⁶). These figures reflect physically realistic accuracy for a genuine 24-hour-ahead forecasting horizon, following correction of an earlier data-leakage issue in which previously reported metrics had inadvertently reflected 1-hour-ahead persistence forecasting rather than true day-ahead prediction. These quantum-refined temperature forecasts are fed directly into an IEC 61215-compliant NOCT thermal derating model, linking prediction quality to theoretical PV output estimation and battery state-of-charge management. Continuous hardware validation on 08 March 2026 confirmed stable operation across all sensing channels. Notably, the 10-parameter quantum correction layer delivers accuracy improvements comparable to a classical neural-network residual-correction layer that would require over 250 trainable parameters — pointing toward a compelling path for parameter-efficient, energy-aware edge AI.
Keywords: Internet of Things, Quantum Machine Learning, Solar Energy Forecasting, Random Forest, Variational Quantum Regressor, ESP32, NASA POWER, Photovoltaic Systems, Edge Computing, Hybrid Classical-Quantum Architecture, Next-Day Temperature Forecasting.