04618ntm a2200337 i 4500 MY-KuUP 20261005145103.0 t||||fr|||| 000 0 ta 261005t20262026my a|||fr|||| 000 0 eng d THE0010810 (Local) Hardback UMPSA eng UMPSA rda FTKEE .A45 2026 r Thesis Amirah Nurhafizah Binti Abu Bakar, author. Artificial neural network-based solar irradiance measurement system applied for photovoltaic modules / Amirah Nurhafizah Binti Abu Bakar Kuantan, Pahang : UMPSA, 2026 © 2026 xiv, 99 pages : illustrations (some color) ; 30 cm. + 1 CD-COM rdacontent text rdamedia unmediated rdacarrier volume rda text file PDF Faculty of Electrical and Electronics Engineering Technology Thesis (Master of Science) -- Universiti Malaysia Pahang – 2026 Includes bibliographical references In recent years, solar photovoltaic power generation has been widely adopted across many countries. To ensure the security and reliability of solar power systems, energy monitoring facilities have been established to promptly address any issues that arise. Solar irradiance is typically measured using conventional equipment such as pyranometers and solarimeters, which are commercially available. However, these traditional sensors and devices often fail to account for real-world factors like dirt accumulation, dust, micro-cracks, and module degradation. This leads to discrepancies between measured and actual irradiance values. Up to recent, numerous studies have been published on solar irradiance prediction. However, it is rare to find research that extensively integrates the development of solar radiation sensors with predictive modeling. A reliable solution that aligns theoretical irradiance measurements with real-world field conditions has yet to be realized. To address this challenge, this study introduces an Artificial Neural Network (ANN)-based solar irradiance measurement system to accurately measure the actual solar irradiance reaching the top surface of solar modules. This system captures current and voltage values, which serve as inputs to the ANN using the nntool in MATLAB. The ANN is designed with four hidden neurons, and the dataset is divided into 80% for training, 10% for testing, and 10% for validation. Comparisons between the ANN method and the Adaptive Neuro-Fuzzy Inference System (ANFIS) method were conducted to validate the ANN-based system's performance using metrics such as correlation coefficient (R), coefficient of determination (R²), root mean square error (RMSE), and mean bias error (MBE). The model with the lowest RMSE and MBE values, along with the highest R and R² values, was selected as the best measurement system for predicting actual solar irradiance. The results from 660mW, 10W and 100W solar module demonstrate that the ANN-based solar irradiance measurement system outperforms ANFIS, achieving R, R², RMSE, and MBE values of 0.9699, 0.9406, 62.2163 W/m2, and -1.5611 W/m2, respectively. The results indicate that the ANN model demonstrated better performance than the ANFIS model, achieving a 6.19% lower RMSE, a 0.87% higher R², a 0.21% higher R, and a 90.7% reduction in bias (MBE). These improvements indicate higher prediction accuracy, reliability, and robustness of the ANN model. The system was further validated using a 50W solar module, proving its compatibility with various solar module specifications and orientations. The proposed ANN-based solar irradiance measurements by demonstrating robust predictive accuracy and generalizability. The system achieves a high correlation coefficient and coefficient of determination, indicating strong agreement between predicted and actual values. Additionally, the RMSE of 62.2163 W/m2 and MBE of -1.5611 W/m2 highlight the model’s ability to maintain both precision and minimal bias in its predictions. The model’s validation across solar modules with varying power ratings (660mW, 10W, 50W and 100W) further emphasizes its adaptability and effectiveness in accommodating diverse specifications and orientations, thereby demonstrating its potential for practical application in dynamic real-world conditions. Faculty of Electrical and Electronics Engineering Technology Dissertations Universities and colleges Dissertations Thesis Dissertations lcc THESIS 0 0 lcc 0 -2 REF 20000 20000 2026-10-05 0 FTKEE .A45 2026 r Thesis T000005648 2026-10-05 14:51:29 1 2026-10-05 THESIS 0 0 lcc 0 -2 REF 20000 20000 2026-10-05 0 CD14063 T000005649 2026-10-05 14:52:03 1 2026-10-05 ACCOM 105667 105673