IEEE Access · 2024 · 23 citations · 39 references
ComputingComputer ScienceMachine LearningRenewable Energy IntegrationRenewable Energy SystemsDeep LearningEnergy Performance IndicatorsEnergy System ModelingNeural NetworksRenewable EnergySolar Photovoltaic SystemsFeedforward Neural NetworksGrid-connected Photovoltaic SystemsInterpretable ModelingInterpretable Neural NetworksLocal ExplanationsModel-agnostic ExplanationRenewable Energy Performance IndicatorsSystem Performance Modeling
Solar power is an important renewable resource in our journey towards a sustainable energy future; however, integrating it with existing grids, especially in dust-prone environments, presents challenges, such as power reduction and financial impact. Regular performance assessment is crucial for identifying issues and maximizing energy production. Therefore, the development of an accurate and reliable predictive model is essential. Such a model should not only predict photovoltaic (PV) system performance but also offer insights into various factors influencing system efficiency. In this regard, this study presents the development of an interpretable deep learning model for the assessment of photovoltaic (PV) system performance. This model focuses on predicting the essential key performance indicator (KPI) performance ratio, which is crucial for PV system evaluation. A feedforward neural network (FFNN) architecture enhanced by a univariate linear regression approach was employed to comprehend the coefficient weights for interpretability. To optimize the model, various optimizers were explored during model training. Furthermore, Local Interpretable Model-agnostic Explanations (LIME) were utilized to determine the influence of specific factors on each prediction made by the FFNN model, enhancing its explainability. The performance of the model was evaluated using standard metrics, such as R-squared (R2)(0.9965), Mean Absolute Error (MAE)(0.0036), Mean Squared Error (MSE)(0.0001), and Root Mean Squared Error (RMSE)(0.0078). The results indicate that the proposed model outperforms conventional deep-learning models, demonstrating promising accuracy and interpretability for PV system performance assessments. By providing insights into the factors affecting PV system performance, our model aims to assist operators and stakeholders in making informed decisions to optimize solar energy utilization.
39
Performance parameters for grid-connected PV systems
Bill Marion, J. Adelstein, K. Boyle et al. · 2005 · 488 citations · Full text
Khaled Obaideen, Maryam Nooman AlMallahi, Abdul Hai Alami et al. · International Journal of Thermofluids · 2021 · 276 citations · Full text
Solar Power Forecasting Using CNN-LSTM Hybrid Model
Su-Chang Lim, Jun‐Ho Huh, Seok-Hoon Hong et al. · Energies · 2022 · 218 citations · Full text
Khaled Obaideen, A.G. Olabi, Yaser Al Swailmeen et al. · Sustainability · 2023 · 217 citations · Full text
Engineering, Environmental Engineering, Environmental Sciences +17